All Activity

“Hi Frank, I had ChatGPT look at our SEO and it has a bunch of recommendations. Can you take care of this for us?” We’re all getting some version of this email from clients and bosses. Responding is fraught with challenges. How do you avoid sounding defensive or dismissive? How do you avoid looking territorial while still explaining that some of these recommendations are generic, flawed, or completely wrong? It’s one thing to know SEO. It’s another to know how to respond tactfully when AI-generated recommendations are suddenly part of the conversation. Resist the urge to respond, ‘ChatGPT is wrong’ It might feel good to tell them the AI output they sent is wrong and that they should leave the SEO to you. But that response usually backfires. It makes you sound defensive, and it shifts the conversation away from SEO and toward whether you’re being territorial. Don’t debate ChatGPT. Show the person who sent the recommendations that you can evaluate AI output objectively and professionally. The first step is acknowledging the effort behind the recommendations before you start evaluating what’s actually useful. Your customers search everywhere. Make sure your brand shows up. The SEO toolkit you know, plus the AI visibility data you need. Start Free Trial Get started with Validate the effort Don’t jump directly into your analysis. Start by thanking them for sending it. Most people forwarding ChatGPT output think they are being helpful. They are trying to contribute ideas, move things forward, or make sure nothing is missed. If your first move is to attack the recommendations, they will hear you attacking the effort. Here’s how we opened a recent client response: “Hi Dr. _______! Thanks for sending this over. There are a few ideas worth taking a look at. I also have some ideas on data we can give the model so it has more context. I’ll follow up with you afterward with more details.” That response does a few important things: It acknowledges the effort behind the recommendations. It signals that you’re evaluating them objectively. And it gives you room to separate the useful ideas from the flawed ones later. You’re not admitting that AI uncovered major issues you missed. You’re showing willingness to review the recommendations professionally before making decisions. Follow up with what’s worth exploring Don’t lead with everything ChatGPT got wrong. Start with the ideas worth exploring first. That shows you evaluated the recommendations objectively instead of dismissing them outright. This is where you demonstrate expertise. Don’t dismiss recommendations simply because they came from AI. Assess whether the underlying observation is valid, whether it matters, and whether it’s worth acting on. For example, I recently reviewed AI-generated feedback on a page our team was working on. Had a client sent this over, an appropriate response would start with something along the lines of: “Thanks for sending this over. I took a look and there is some room to get some more Philadelphia-relevant content/language into this page while keeping it natural. I’ve assigned this to one of our copywriters to get started.” Get the newsletter search marketers rely on. See terms. Let the sender come to the conclusion that ChatGPT is wrong Once you’ve acknowledged the recommendations worth exploring, you can start addressing the weaker ones. The key is to walk stakeholders through the reasoning rather than simply declaring the AI output is wrong. For example, we received an AI-generated analysis from a plastic surgeon client claiming that several competitors had “focused their SEO” around a single procedure: “Hi Dr. ___________, Positioning you as the surgeon in your market for a specific procedure goes beyond SEO. This is a fundamental aspect of branding and positioning that could not only drive better user signals, resulting in better rankings, but higher conversion rates as well. I would note that if you visit these websites, however, you’ll see that they rank well for facelift queries even though they list many other procedures. I can’t figure out why the model is claiming that their SEO is focused on facelift. They are producing content beyond that procedure, as well: So if you decide to go all-in on positioning yourself around a specific procedure, it doesn’t mean we can’t list other procedures on the website, nor does it mean we’d be limited to writing only about that procedure. It would largely direct efforts on social media, outdoor, and areas outside of SEO and the website.” Notice what this approach does. It: Acknowledges the valid strategic point behind the recommendation. Introduces contradictory evidence calmly. Allows the stakeholder to recognize the flaw in the AI’s reasoning independently. That is far more persuasive than simply saying “ChatGPT is wrong.” Focus on improving the analysis, not debating the output At some point, you need to explain what is really happening: AI outputs are only as good as the prompt and context they are given. In this case, our client did not provide any context, data, or guidance. He simply asked the model for SEO recommendations for his website. Continuing the client email: “…the model is recommending we add procedure pages in excess of 3,000+ words: Luckily, we already have all of these pages up, though our word counts do not exceed 3000. I checked this against the top-ranking results for these queries and found that almost all have word counts that are much lower than this, which reflects my experience that raw word count does not drive rankings: I think we should rerun this analysis and make a few changes to the prompt, including asking it to ignore word count. We should also ask the model to analyze these pages against ours and point out subtopics they cover that our pages have missed, entities they include that we don’t, and how the information density of our content compares to theirs. What do you think?” Notice the shift in the conversation. You’re not arguing about whether ChatGPT is “right” or “wrong.” You’re improving the quality of the analysis itself. This is a much more productive conversation that positions you as collaborative, analytical, and confident in your expertise instead of defensive about it. See the complete picture of your search visibility. Track, optimize, and win in Google and AI search from one platform. Start Free Trial Get started with These emails aren’t going away. Learn how to answer them. You’re going to get more and more of these emails from clients, executives, and internal stakeholders. Learning how to respond to them effectively will become an increasingly important part of SEO and marketing leadership. The challenge isn’t just evaluating AI-generated recommendations. It’s doing so in a way that: Keeps stakeholders engaged. Reinforces your expertise. Doesn’t consume unnecessary time and energy. The next time you feel tempted to send AI-generated recommendations to your accountant, doctor, or IT department, remember what it feels like to be on the receiving end of them. View the full article

Michaels is expanding its party supply and celebration offerings. In September 2025, the arts and crafts retailer introduced The Party Shop at Michaels, an in-store shopping experience that brought party supplies, balloons, and other celebration essentials to its shelves. This year, its product selection will grow even further. In a May 13 press release, Michaels announced it is expanding its in-store party supply assortment and introducing new in-store experiences, with plans to add nearly 600 new products to its shelves throughout 2026. Michaels isn’t the only unexpected retailer with a party supply aisle. Last month, Staples announced it was getting into the party business with help from Party City. The office supply retailer shared plans to add Party City at Staples shop-in-shops to more than 700 of its stores. Party City filed for bankruptcy for a second time in December 2024 and announced it would close all of its stores. Its departure left many shoppers without a go-to retailer for party supplies and balloons. In the wake of its store closure, retailers like Staples and Michaels are filling the gap. Michaels offers new ways to prepare for your next party In the past year, Michaels has expanded its party supplies and balloon offerings to over 4,500 products. The retailer isn’t slowing down—it plans to add around 600 new products by the end of 2026, bringing even more celebration finds to shelves. New products include piñatas, expanded licensed essentials featuring characters like Hello Kitty and Bluey, and year-round entertaining products. Michaels is also introducing new in-store experiences. Beginning this month, the craft retailer is rolling out the following in-store DIY customization bars across North America: The Favor Bar: Mix and match items to build custom party favors. The Candy Bar: Fill favor bags or create dessert displays with an assortment of sweets. The DIY Banner Bar: Create personalized felt banners with interchangeable numbers, letters, and icons. And for those who need gift wrap, Michaels will have an assortment of gift wrap and bags, bows, tags, and tissue paper—customers can mix and match five items for $5. “At Michaels, we believe the joy of celebrating should begin the moment you start planning,” David Boone, CEO of Michaels, said in a statement. Shop-in-shops are the latest way retailers are expanding The Party Shop at Michaels isn’t the only in-store shopping concept that the craft retailer offers. Last year, Michaels also welcomed The Knit & Sew Shop. The shop features Joann and Michaels-branded sewing and crafting essentials like yarn, thread, and fabric. In 2025, the arts and fabrics chain Joann Inc went bankrupt and closed all its remaining stores. In June 2025, Michaels acquired Joann’s intellectual property and private label brands. Shop-in-shops like The Knit & Sew Shop and The Party Shop at Michaels allow retailers to expand their offerings and appeal to a wider customer base without expanding their physical footprint by opening new stores. Both Michaels and Staples are privately held companies after formerly being publicly traded. The Michaels Companies was taken private in 2021 by Apollo Global Management. Staples Inc was bought in 2017 by Sycamore Partners. View the full article

European nations swing behind idea discussed with Riyadh to model agreement on 1970s Helsinki processView the full article

Presidents Xi Jinping and Donald The President started a crucial series of meetings in Beijing on Thursday in a U.S.-China summit where stability in the relationship is the main goal of the two days of discussions. The White House and Chinese state media said the leaders concluded their meeting Thursday morning after about two hours. The President is expected to leave just after midday Friday after a final private meeting with Xi. But few breakthroughs are expected on divisive issues ranging from the Iran war, trade, technology and Taiwan. The President hopes to focus the summit talks on trade and deals for China to buy more agricultural products and passenger planes, setting up a board to address their differences and avoid a repeat of the trade war ignited last year after The President’s tariff hikes. In their closed-door meeting, Xi told The President that if Taiwan is handled well, U.S.-China relations “will enjoy overall stability.” If not, the two countries risk “clashes and even conflicts, putting the entire relationship in great jeopardy,” Xi said, according to China’s official Xinhua News Agency. The President in December authorized an $11 billion arms package for Taiwan, a self-governed island that Beijing claims as its own territory. The U.S. has not yet moved forward with delivery. Xi said China’s door of opening to U.S. business will only open wider, he told American corporate leaders who accompanied The President. The U.S. president said the business leaders all respect and value China and he encourages them to expand cooperation with China, Xinhua reported. The war with Iran is also likely to be a key topic. Ahead of the meetings, The President hoped China would use its considerable leverage to prod Iran to agree to U.S. terms to end the two-month old war or reopen the critical Strait of Hormuz, but he has tempered those calls ahead of the summit. —Associated Press View the full article

Grasping the benefits of customer loyalty is essential for your business’s success. Loyal customers not just stay longer but likewise spend considerably more than new ones, which directly impacts your profits. They often share their positive experiences, enhancing your brand’s visibility through word-of-mouth. Furthermore, these customers provide insights that can refine your offerings and strengthen your market position. Exploring these advantages can lead to strategic decisions that nurture long-term growth and sustainability. Key Takeaways Increased customer retention can significantly boost profits, with a mere 5% increase elevating profits by 25% to 95%. Loyal customers spend 67% more than new customers, enhancing revenue stability and profit margins. Word-of-mouth referrals from satisfied customers can reduce acquisition costs by up to 50%. Strong customer loyalty fosters brand advocacy, as loyal customers are more likely to recommend your brand. Engaging loyal customers provides valuable insights and feedback, informing business improvements and marketing strategies. Increased Customer Retention When businesses prioritize customer retention, they not merely save on acquisition costs but in addition improve their overall profitability. Why is customer loyalty important? Retaining existing customers is five times cheaper than acquiring new ones, making it a cost-effective strategy. A mere 5% increase in customer retention can boost profits by 25% to 95%. Loyal customers are also 50% more likely to try new products, creating broader sales opportunities. Moreover, customers with strong loyalty exhibit a 14 times higher likelihood of making repeat purchases compared to new customers. Implementing loyalty programs can improve retention rates considerably, with 77% of consumers stating they’re more likely to stay with brands that offer such programs. These benefits of customer loyalty demonstrate its crucial role in business success. Enhanced Customer Lifetime Value Improved customer lifetime value (CLV) is a vital metric that reflects the total worth a customer brings to your business over the duration of their relationship. Focusing on customer loyalty can greatly improve this value, as loyal customers are worth 306% more than non-loyal ones. Here are three key factors to contemplate: Retaining just 5% of your customers can increase profits by 25% to 95%, making loyalty essential for maximizing CLV. Loyal customers usually spend 67% more than new customers, boosting your average order value. Long-term relationships nurtured by loyalty programs create stability in sales, ensuring a steady revenue stream. Boosted Word-of-Mouth Marketing When you create loyal customers, you’re tapping into a formidable marketing tool: word-of-mouth recommendations. These satisfied customers aren’t just 77% more likely to share their positive experiences, but their endorsements carry more weight than traditional ads, leading to greater trust. Amplified Brand Recommendations How can customer loyalty greatly improve brand recommendations? When customers are loyal, they become enthusiastic advocates for your brand. This enthusiasm translates into recommendations that profoundly impact your business. Here are three key benefits: Increased Referrals: Loyal customers are 77% more likely to recommend your brand after a positive experience, boosting your word-of-mouth marketing efforts. Trusted Endorsements: Approximately 47% of consumers show loyalty by recommending brands they trust, and these endorsements carry more weight than traditional ads. Cost Efficiency: Word-of-mouth referrals from satisfied loyal customers can lower your acquisition costs, as authentic endorsements resonate better with potential buyers. Trustworthy Organic Endorsements Building on the enthusiasm generated by loyal customers, trustworthy organic endorsements play a pivotal role in enhancing word-of-mouth marketing. When loyal customers have a positive experience, they’re 77% more likely to recommend your brand to friends. These recommendations carry 92% more credibility than traditional advertising, making loyal customers influential advocates. Approximately 47% of consumers share their positive experiences, driving organic growth through authentic endorsements. Additionally, word-of-mouth referrals can lower your customer acquisition costs by up to 50%. Brands with strong customer loyalty enjoy increased visibility, as satisfied customers often amplify their positive experiences on social media. Improved Brand Trust and Advocacy When you build loyalty with your customers, you improve their emotional connection to your brand, which can lead to increased trust and advocacy. Loyal customers are more likely to recommend your products to their friends, driving valuable word-of-mouth referrals that traditional advertising can’t match. This not just boosts your brand visibility but additionally nurtures long-term relationships that keep your customers engaged and less likely to switch to competitors. Enhanced Emotional Connections Emotional connections play a crucial role in enhancing brand trust and advocacy, as they directly influence customer loyalty. When you cultivate these connections, you can expect the following benefits: Increased Loyalty: Emotional ties can lead to a 26% rise in true loyalty, translating into stronger advocacy for your brand. Higher Spending: Loyal customers are 50% more likely to try new products and spend 31% more than new customers, demonstrating that trust encourages greater purchasing behavior. Reduced Price Sensitivity: Brands with emotional connections can mitigate price sensitivity, as loyal customers prioritize trust over competitive pricing, making them less likely to switch for cheaper options. Increased Word-of-Mouth Referrals Customer loyalty greatly improves word-of-mouth referrals, which can profoundly impact your brand’s reach and reputation. When customers have positive experiences, they become 77% more likely to recommend your brand to friends, considerably enhancing your visibility. In addition, about 47% of loyal customers are inclined to share their experiences, acting as trusted advocates for your brand. This organic word-of-mouth marketing can streamline customer acquisition, lowering your marketing costs and allowing you to allocate resources more effectively. Recommendations from satisfied loyal customers carry more weight than traditional advertisements, leading to higher conversion rates among potential new customers. Higher Profit Margins Loyalty among your customers can greatly impact your business’s profit margins, as they tend to spend 67% more than new customers. This loyalty leads to several key advantages: Cost Efficiency: Retaining existing customers is five times cheaper than acquiring new ones, which cuts down on marketing costs and boosts profit margins. Increased Retention: A mere 5% increase in customer retention can elevate profits by 25% to 95%, underscoring the financial benefits of loyalty. Pricing Flexibility: Loyal customers are less sensitive to price changes, enabling you to implement higher pricing strategies without losing sales, further improving profit margins. Valuable Customer Insights How can valuable insights from your customers drive business success? By leveraging loyalty programs, you gain rich data on customer preferences and behaviors. This information helps you customize your offerings and marketing strategies effectively. For instance, analyzing loyalty member data can reveal which products boost customer engagement and satisfaction. Here’s a quick overview: Insight Type Benefit Example Customer Preferences Customized marketing strategies Personalized email campaigns Product Development Improved product offerings New flavors based on feedback Customer Feedback Enhanced customer experiences Service improvements Using this data, you can track shopping frequency and spending habits, leading to better-targeted promotions and increased revenue. Competitive Advantage in the Market In today’s competitive environment, businesses must leverage every advantage to stand out. Customer loyalty can provide a significant competitive edge, helping you secure and grow your market share. Here are three key benefits: Brand Stability: Strong loyalty programs differentiate your brand, making customers less likely to switch to competitors, which improves stability in a crowded market. Innovation Acceptance: Loyal customers are 50% more likely to try new products, allowing you to introduce innovations that will be well-received. Mitigated Price Competition: Loyal customers prioritize relationships over price, reducing sensitivity to price changes and easing competitive pressures. Frequently Asked Questions Why Is Customer Loyalty Important for a Business? Customer loyalty is important for a business since it markedly reduces costs associated with acquiring new customers. When you retain existing customers, you save money and increase profitability. Loyal customers tend to spend more over time and are likely to recommend your brand to others, enhancing your reputation. Furthermore, even a small increase in customer retention can lead to substantial profit growth, demonstrating how loyalty directly impacts your revenue and long-term success. What Are the Key Benefits of Customer Loyalty for a Business and How Do These Advantages Contribute to Its Long-Term Success and Profitability? Customer loyalty brings numerous advantages that greatly contribute to long-term success and profitability. For instance, loyal customers tend to spend more, often 67% more than newcomers, enhancing revenue. Furthermore, retaining existing customers is usually five times cheaper than acquiring new ones, which improves cost efficiency. In addition, satisfied customers are likely to recommend your brand, increasing brand awareness through word-of-mouth. As a result, nurturing loyalty can lead to substantial profit growth and sustainable business development. What Are the 4 C’s of Customer Loyalty? The 4 C’s of customer loyalty are Commitment, Consistency, Communication, and Community. Commitment reflects the emotional connection between you and the brand, driving repeat purchases. Consistency guarantees you receive reliable experiences, nurturing trust. Effective Communication makes you feel valued, improving your satisfaction. Finally, Community builds a network of loyal customers who engage with each other, amplifying word-of-mouth recommendations. Together, these elements create strong relationships that improve customer loyalty and overall brand success. What Are the 3 R’s of Customer Loyalty? The 3 R’s of customer loyalty are Retention, Referral, and Revenue. Retention focuses on keeping existing customers, as it’s markedly cheaper than acquiring new ones. A small increase in retention can lead to substantial profit gains. Referral emphasizes that satisfied customers are likely to recommend your brand, enhancing word-of-mouth marketing. Finally, Revenue highlights that loyal customers typically spend more, which boosts your overall profitability and contributes to a higher Customer Lifetime Value. Conclusion In summary, nurturing customer loyalty is crucial for your business’s growth and sustainability. By focusing on increasing retention, enhancing customer lifetime value, and encouraging word-of-mouth marketing, you position your brand for long-term success. Loyal customers not just contribute to higher profit margins but likewise provide valuable insights and create a competitive advantage in the market. Prioritizing these aspects can greatly improve your overall business performance and help you build a strong, reputable brand in your industry. Image via Google Gemini and ArtSmart This article, "7 Key Benefits of Customer Loyalty for Your Business" was first published on Small Business Trends View the full article

When your brand disappears from ChatGPT or Perplexity, the fix isn't more content. It's diagnosing which layer broke down. The post Stop Treating AI Visibility As One Problem. It’s Actually Three, On Three Different Layers appeared first on Search Engine Journal. View the full article

We may earn a commission from links on this page. Deal pricing and availability subject to change after time of publication. Samsung’s 27-inch Odyssey G5 (G51F) gaming monitor has dropped to $159.99 on Amazon, which is the lowest price it has reached so far, according to price trackers. That’s a noticeable discount from its usual $249.99 price, and it makes a lot more sense now for anyone ready to move on from a basic 1080p setup without jumping into the much higher cost of OLED displays. It’s a flat panel (unlike Samsung’s many curved gaming displays), with a matte coating that helps minimize glare in brighter rooms, and comes with a stand that supports height, tilt, and pivot adjustments (something many budget gaming monitors skip entirely). Samsung 27" Odyssey G5 (G51F) Gaming Monitor $159.99 at Amazon $249.99 Save $90.00 Get Deal Get Deal $159.99 at Amazon $249.99 Save $90.00 The G51F’s combination of 180Hz refresh rate, 1ms response time, and AMD FreeSync support makes fast-paced games look smoother and feel more responsive than they do on standard 60Hz displays, especially in shooters, racing games, and competitive multiplayer titles. The VA panel also helps the monitor deliver deeper blacks and stronger contrast than many IPS alternatives in this price range, so darker games and movies tend to look less gray and washed out. That said, while HDR10 support is included, buyers should keep expectations realistic—with 300 nits of brightness, this is more of a basic HDR experience than the kind of dramatic HDR you get from higher-end Mini LED or OLED displays. Outside of gaming, the Odyssey G5 works reasonably well as a general-purpose monitor too. The sharper 1440p resolution makes multitasking easier, and the extra screen space helps when editing photos, managing spreadsheets, or keeping multiple windows open. Connectivity is decent as well, with HDMI, DisplayPort, and USB support for accessories and peripherals. That said, like many VA panels, it can show some motion smearing in darker scenes, and people who mainly play competitive esports games may still prefer faster IPS or OLED options. Still, for under $160, this makes for a practical upgrade for someone who wants sharper visuals, smoother gameplay, and a more versatile display without overspending. Our Best Editor-Vetted Tech Deals Right Now Apple AirPods Pro 3 Noise Cancelling Heart Rate Wireless Earbuds — $229.00 (List Price $249.00) Apple Watch Series 11 [GPS 46mm] Smartwatch with Jet Black Aluminum Case with Black Sport Band - M/L. Sleep Score, Fitness Tracker, Health Monitoring, Always-On Display, Water Resistant — $329.00 (List Price $429.00) Apple iPad 11" A16 128GB Wi-Fi Tablet (Silver, 2025) — $319.99 (List Price $349.00) Shark AV2501AE AI XL Hepa- Safe Self-Emptying Base Robot Vacuum — $299.99 (List Price $649.99) Dell 15 DC15250 (Intel Core i7 13th Gen, 512GB SSD, 8GB RAM, Touch Display) — $599.99 (List Price $839.99) Deals are selected by our commerce team View the full article

LinkedIn has always been a key driver for B2B discovery, but over the past few years, a new layer of upper-funnel clout has developed: the platform’s influence on AI search citations. LLMs are increasingly influential in how B2B buyers discover products and services, and LinkedIn has become a top source of this information. This means that if your brand effectively optimizes its LinkedIn presence and content flow for AI search ingestion, you’ll likely get a corresponding bump in AEO-based discovery. In our work with B2B clients (mostly of the high-growth SaaS variety), we’ve divided this LinkedIn AEO initiative into three segments: Optimize earned media. Feed LLMs strategic content. Invest in post-engagement that strengthens LLM signals. Here’s how to approach each segment and the outcomes you can expect. 1. Optimize earned media: website, company pages, and high-profile employee pages If you need reasons to keep your website optimized and your LinkedIn pages (both your company page and the pages of your high-profile employees, such as content contributors and thought leaders) up to date, here you go: Doing so feeds LLMs signals that your brand is trustworthy and an authentic source of information. Much like Google adheres to E-E-A-T for traditional SEO, LLMs pull signals from a brands’ earned media to gauge credibility and trustworthiness. Content published on behalf of a brand by its employees and leaders can also contribute to the brand’s reputation, provided those authors are optimizing their owned media. On websites Make sure your business address, contact information, product descriptions, about pages, and author profile pages are fully built out with good, accurate information. On LinkedIn company pages Pay attention to top-level positioning, your “About” section, and the products and services you offer, providing good, detailed descriptions for each. This may seem basic, but it’s common for companies to go for excessively long periods without updating their LinkedIn pages, beyond just posting. Take 30 seconds to gauge whether your page is truly up to date or missing messaging that’s integral to appearing in relevant LLM prompts. (For example, if your products and services are particularly relevant to a specific industry, call out that industry in your intro text.) One more thing: Make sure your company’s executives and thought leaders also have your company and positioning reflected in their profiles. Better yet, they should be posting on behalf of the company if they’re willing to use their profile on its behalf – that’s just more material telling the LLMs that your company is a real, authentic, trustworthy source of expertise. Your customers search everywhere. Make sure your brand shows up. The SEO toolkit you know, plus the AI visibility data you need. Start Free Trial Get started with 2. Feed the LLMs strategic content Just as a 100-word blog would be a huge outlier to move the needle in organic discovery, meatier content on LinkedIn has proven more influential for AEO visibility (according to recent research our agency’s LinkedIn rep shared with us). Specifically, 800 to 1,200 words of high-quality, original, differentiated content seems to be a great target for driving AEO mentions. LinkedIn articles and newsletters are perfect forums for this length, since users open them expecting deep dives and won’t instinctively bounce like a Facebook user clicking “…more” on a post only to see a mass of text below. Yes, carousels and videos are great for building engagement, and there’s every reason to embed them in newsletters and articles. But early signs are that LLMs really, really like good, richly written content. Dig deeper: LinkedIn Ads on a budget: How one playbook drove sub-$10 CPL Get the newsletter search marketers rely on. See terms. 3. Invest in building post-engagement More research from our LinkedIn rep: LinkedIn posts with at least 10 quality comments and/or 60 reactions are particularly influential for LLMs. That makes sense, as social proof is a strong signal of authority, and it’s important to note that achieving this level of engagement doesn’t require a ton of added budget. Yes, you can boost company posts and use Thought Leader Ads (TLAs) and follower ads to build bigger user bases. I almost always recommend brands test TLAs when having employees do the work of putting up good, relevant content It’s a good practice to do this anyway — LLMs or not — for posts that get good organic traction and effectively speak to a company’s products, services, or positioning. Our rep didn’t have any precise data that indicated a correlation between TLAs/boosted posts and greater visibility on LLMs. However, as TLAs and boosted posts are essentially promoted organic posts, they serve as a foundation for stronger organic traction. Another LinkedIn threshold to note for AEO is that engagement from profiles with less than 3,000 followers (again, this is from our LinkedIn rep) tends to carry more clout with LLMs because those profiles are seen as relatively authoritative and trustworthy. If you have any employees (including executives) who are over that threshold, empower them to post on behalf of the business by helping them share insights, proprietary data, and any effective tests or methodologies that have driven good results. (While a lot of companies prefer not to tip their hand on the latter, doing so is a great way to build a broad reputation for expertise.) Don’t stop at employees, either: Consider follower ads to build your company’s follower base, and see if you can form partnerships with verified industry experts (guest blogs and video interviews are great for this) who will amplify your brand’s content. Just ensure that content follows the thought leadership thread of the section above; overly promotional content and straight-up brand messaging won’t get much traction with any audience, machine, or human. Dig deeper: LinkedIn’s new playbook taps creators as the future of B2B marketing AI search is expanding LinkedIn’s influence in B2B AEO must now be a careful consideration in your approach to every channel, including Reddit and YouTube. If you’re in B2B and sticking close to the in-platform data you see from LinkedIn, zoom out and carve out some resources to address the initiatives above. The impact of AEO is hard to measure, but it’s only growing as B2B users flock to the LLMs. View the full article

The World Cup final will feature a star-studded halftime show headlined by Madonna, Shakira and boy-band BTS. FIFA has announced that, for the first time, the final at the MetLife Stadium in New Jersey on July 19 will include a Super Bowl-style concert. The governing body said the show would support the FIFA Global Citizen Education Fund, which is raising $100 million to help children access education and soccer. FIFA president Gianni Infantino said it would bring together “music and football on the biggest stage in sport for a very special cause.” “Every child should have the opportunity to dream, and together we can help make that possible,” he posted on Instagram. The show will be curated by Coldplay’s Chris Martin. The Super Bowl is famed for its halftime show — attracting the world’s biggest stars for spectacular performances. This year featured Puerto Rican artist Bad Bunny. Previous headliners included Michael Jackson, Paul McCartney, the Rolling Stones, Madonna, Prince, Bruce Springsteen and Rhianna. But halftime shows are not so commonplace in soccer, with events such as the Champions League final featuring a pre-match concert. This year will see the Killers headline European club soccer’s biggest game between Paris Saint-Germain and Arsenal in Budapest. FIFA describes its halftime show as “a singular moment at the intersection of sport, culture and purpose, broadcast live around the world.” This year’s World Cup is co-hosted by the United States, Canada and Mexico and runs through June and July. AP soccer: https://apnews.com/hub/soccer View the full article

On Wednesday, Cisco Systems announced impressive quarterly earnings alongside nearly 4,000 job cuts. The dichotomy stemmed from the hardware and networking company’s embrace of a rapidly growing trend in tech: openly admitting that layoffs are due to AI adoption rather than poor performance. “The companies that will win in the AI era will be those with focus, urgency, and the discipline to continuously shift investment toward the areas where demand and long-term value creation are strongest,” Cisco CEO Chuck Robbins told employees in a publicly shared email. “I’m confident Cisco will be one of those winners. This means making hard decisions—about where we invest, how we’re organized, and how our cost structure reflects the opportunity in front of us.” With his announcement, Robbins follows in the footsteps of tech leaders including Block CEO Jack Dorsey and Snap CEO Evan Spiegel, who made similar moves this year. Robbins emphasized that the company will further invest in employees’ AI use throughout their jobs. Meanwhile, employees will start getting notifications if they’ve been laid off on Thursday. Cisco says the job cuts make up less than 5% of its total workforce. Shares of Cisco Systems Inc. (Nasdaq: CSCO) were up more than 16% on Thursday morning. The stock had already been trading at record highs this month. How did Cisco perform during its third quarter? Cisco reported $15.8 billion in revenue for the quarter ending on April 25. That figure represents a 12% jump year-over-year (YOY) and beats Wall Street’s predicted $15.56 billion, according to consensus estimates cited by CNBC. The company also surpassed expectations of $1.04 earnings per share with $1.06 adjusted. In a post-earnings call, Robbins highlighted AI-centric business with companies like Nexus and Nvidia, as well as a significant increase in revenue from AI. For instance, this quarter, Cisco shared plans to expand its secure AI factory with Nvidia. Cisco’s product revenue rose 17%, something Robbins attributes to “robust demand for our AI infrastructure and campus networking solutions.” Cisco expects its revenue to reach $16.7 billion to $16.9 billion in quarter four and $62.8 billion to $63 billion for fiscal year 2026. In comparison, it saw $56.7 billion in revenue for fiscal year 2025. View the full article

We may earn a commission from links on this page. Deal pricing and availability subject to change after time of publication. OLED gaming monitors have become much easier to recommend over the last year, mostly because prices have started dropping below the $500 mark. LG’s 27GX704A-B UltraGear OLED is one of the better options in that category right now, with excellent motion handling, extremely low input lag, and the glossy WOLED panel that gives games a cleaner, more contrast-heavy look than many matte alternatives. It’s also down to $477.99 on Amazon, a 40% discount from its usual $799.99 price. LG 27GX704A-B Ultragear OLED Gaming Monitor $477.99 at Amazon $799.99 Save $322.00 Get Deal Get Deal $477.99 at Amazon $799.99 Save $322.00 At this price, the LG 27GX704A-B lands much closer to premium IPS gaming monitors while still offering the contrast and motion performance OLED panels are known for. This is a 1440p display with a 240Hz refresh rate, so it’s clearly aimed at PC gamers who care more about responsiveness and motion clarity than pushing full 4K resolution. Competitive games like Valorant, Apex Legends, and Call of Duty benefit the most here because the near-instant response time keeps motion looking unusually sharp, even during fast camera movement or flicks across the screen. The panel itself uses LG’s newer third-generation WOLED technology and supports both FreeSync Premium Pro and G-SYNC compatibility, so screen tearing is less of an issue regardless of whether you use an AMD or NVIDIA graphics card. One of the more noticeable differences between this model and LG’s earlier UltraGear OLED displays is the glossy screen coating—instead of the matte finish found on many gaming monitors, this panel looks clearer and a bit punchier in darker rooms because the coating doesn’t soften the image as much. Blacks look genuinely deep, HDR highlights stand out nicely, and games with darker environments benefit a lot from the OLED panel’s per-pixel lighting. The downside is that reflections become much more noticeable if your setup sits near a sunny window or a bright overhead light, and VRR flicker can appear when frame rates bounce around heavily in darker scenes. The LG 27GX704A-B is also positioned as an entry-level OLED gaming monitor, so you won’t get premium features like DisplayPort 2.1, a built-in KVM switch, or an especially advanced USB hub beyond two basic USB-A ports. Our Best Editor-Vetted Tech Deals Right Now Apple AirPods Pro 3 Noise Cancelling Heart Rate Wireless Earbuds — $229.00 (List Price $249.00) Apple Watch Series 11 [GPS 46mm] Smartwatch with Jet Black Aluminum Case with Black Sport Band - M/L. Sleep Score, Fitness Tracker, Health Monitoring, Always-On Display, Water Resistant — $329.00 (List Price $429.00) Apple iPad 11" A16 128GB Wi-Fi Tablet (Silver, 2025) — $319.99 (List Price $349.00) Shark AV2501AE AI XL Hepa- Safe Self-Emptying Base Robot Vacuum — $299.99 (List Price $649.99) Dell 15 DC15250 (Intel Core i7 13th Gen, 512GB SSD, 8GB RAM, Touch Display) — $599.99 (List Price $839.99) Deals are selected by our commerce team View the full article

Today is an important day in the 2026 IPO landscape: Cerebras Systems Inc. is making its much-anticipated market debut. While not a household name like Nvidia, Intel, or TSMC, Cerebras is a chipmaker that is rapidly becoming a critical player in the AI semiconductor space. And investors will be casting a keen eye on how its stock performs in the early days of trading, looking for hints about how other, even more anticipated AI-related listings may play out later this year. Here’s what you need to know about Cerebras and its initial public offering: What is Cerebras Systems? Cerebras Systems is an AI semiconductor company headquartered in Sunnyvale, California. It was founded in 2015 by Andrew Feldman, Gary Lauterbach, Michael James, Sean Lie, and Jean-Philippe Fricker. Feldman is the company’s CEO. The company specializes in making the largest—quite literally—computer chips in the world, chips that are optimized for running AI tasks. While most computer chips are made from large wafers that are then divided to make smaller, individual chips, a single Cerebras chip is the entire wafer. As Fast Company previously reported when it named Cerebras one of the most innovative AI companies of 2026, the large size of its chips means they can perform AI tasks much more quickly—up to 70 times faster than the GPUs that many AI systems run on today. “The large square chip packs a lot of processing power and memory on one piece of silicon, so almost no time is wasted routing data between separate chips,” Fast Company’s Mark Sullivan previously noted. “That makes it highly effective at processing data from commercial AI applications that require massive throughput and very fast response times.” Cerebras’s customers include pharmaceutical companies like AstraZeneca and GlaxoSmithKline, as well as tech firms like G42, IBM, Meta, Mistral, Notion, and Perplexity. Most recently, Cerebras inked a $20 billion deal with ChatGPT maker OpenAI. When is Cerebras Systems’ IPO? Cerebras Systems priced its shares on Wednesday. It is expected to list on Thursday, May 14, 2026. What is Cerebras Systems’ stock ticker? Cerebras Systems’ shares will trade under the stock ticker “CBRS.” The stock will trade on the Nasdaq Global Select Market. What is the IPO share price of CBRS? The initial public offering price for CBRS shares is $185 per share. This final IPO price is remarkably higher than the IPO share price Cerebras said it would pursue just a few weeks earlier. On May 4, the company announced it would initiate the road show for its upcoming IPO. At that time, Cerebras said that the initial public offering price was expected to be between $115 to $125 per share. While it is not uncommon for a company to tweak its IPO price in the days leading up to the actual IPO, the final $185 IPO share price is around 60% higher than the low end of the original range. This suggests that demand for shares was much greater than initially anticipated. How many CBRS shares are available in its IPO? Upon its IPO listing, Cerebras Systems made 30 million shares of its Class A common stock available. The company’s underwriters, which include Morgan Stanley, Citigroup, Barclays, and UBS Investment Bank, also have a 30-day option to buy an additional 4.5 million shares. How much did Cerebras Systems raise in its IPO? Selling 30 million shares at $185 each means Cerebras raised $5.5 billion in its IPO. As noted by CNBC, that makes this offering one of the largest U.S. tech IPOs in recent memory. It puts Cerebras above the $3.8 billion that Snowflake raised in its 2020 IPO, and behind the roughly $8 billion Uber raised in its 2019 IPO. How much is Cerebras Systems worth? At its IPO price, Cerebras is now valued at around $56.4 billion, according to CNBC. 2026 is shaping up to be the year of AI IPOs Given all the hype and hope around AI, it’s little surprise Cerebras’s IPO shares went for significantly higher than the company had originally forecast. And the successful IPO also bodes well for other AI companies that are likely to go public this year. Two of the most anticipated AI-related IPOs of 2026 include Claude maker Anthropic and ChatGPT maker OpenAI. Current rumblings point to Anthropic debuting first, followed by OpenAI by the end of the year. Of course, AI companies aren’t the only tech firms expected to go public in 2026. Another big tech company that will likely IPO this year, perhaps as soon as this summer, is Elon Musk’s SpaceX. Taken all together, 2026 could be one of the biggest years on record when it comes to the total valuation of all tech IPOs scheduled to go public. View the full article

When OpenAI switched default models on March 4, the number of websites cited per response dropped by a fifth, and never recovered. But the citation drop is only part of the story. We also reverse-engineered ChatGPT’s internal browsing tools, ran a honeypot experiment, reconstructed its system prompt, and released a new version of our ChatGPT Search Capture plugin. What happened On March 4, ChatGPT switched its default model from GPT-4o/5.2 to GPT-5.3 Instant. The result: the average number of unique domains cited per response dropped from 19 to 15, a decline of more than 20%. Unique URLs per response followed the same trajectory, falling from 24 to 19. We tracked 400 daily prompts over 14 weeks, using monitoring data provided by Meteoria. Why we care ChatGPT has 900 million weekly active users. The citation surface in each response hasn’t changed, but fewer websites are sharing it. Same pie, fewer slices. This likely reflects a structural shift toward higher-authority sources, but it also means fewer winners overall. Sites that don’t make the cut are losing visibility that was previously within reach. The Bigfoot Effect We named this phenomenon after the “Bigfoot update” (identified by Dr. Peter J. Meyers of Moz in 2012), when Google would sometimes let a single domain occupy the entire first page of results. ChatGPT now retrieves fewer domains per response, but the URL-to-domain ratio has remained stable at 1.26. Crawl depth per domain hasn’t changed. What has changed is how many distinct websites get a seat at the table. GPT-5.4 Thinking amplifies the concentration further. The model uses “site:” operators to restrict searches to trusted domains and distributes its queries across often more than 10 “fan-out queries” per response, each targeting a specific source. Independent log analysis by Jérôme Salomon (Oncrawl) confirms the trend. ChatGPT-User bot crawl volume has settled at a lower level since the switch to 5.3. Some pages simply aren’t being crawled anymore. The cause goes beyond model updates: more than 90% of ChatGPT’s weekly users are on the free plan, and the default experience triggers fewer web searches, uses fewer queries, and produces fewer citations. How ChatGPT Search actually works Our study also includes a full reverse engineering of ChatGPT’s internal search system, called web.run. Before 5.3, the model sent compact text commands separated by pipes (fast|query|recency). After 5.3, it sends structured JSON objects with typed parameters. This isn’t just a format change. It reflects a different architecture in how the model formulates and distributes its web operations. The web tool now supports 12 operations, up from 4 (plus a separate widget system called genui). These include: search_query open find click screenshot product_query Specialized widgets for sports, finance, weather, and more. GPT-5.4 can chain 5 to more than 10 rounds of search per response, refining queries based on previous results. GPT-5.3 Instant typically runs 2 or 3. Google’s fingerprints are still visible: Google tracking markers (strlid) appear in product URLs, and SearchAPI ID-to-token matches reveal the backend’s reliance on third-party search providers — and Google behind the scenes. A new type of fan-out for product queries We uncovered a previously undocumented fan-out type: browse_rewritten_queries. It appears exclusively on product queries, on 5.4 Instant, and is visible in conversation code. When a user asks something like [best 3D printer to buy in 2026] ChatGPT first runs a single rewrite fan-out to build the full list of candidate products. Then it launches a separate shopping fan-out for each individual product, fetching specs, reviews, and pricing one by one. Before 5.3, product searches were bundled into a single call. Each product now gets its own dedicated retrieval command. ChatGPT-User is the retrieval agent Our honeypot experiment confirmed an important detail. When ChatGPT browses the web following a search during a conversation, the ChatGPT-User crawler — not OAI-SearchBot — fetches the page content. OpenAI describes OAI-SearchBot as the agent that builds ChatGPT’s search index, but in practice, the model relies on third-party scraping APIs for search results, then sends ChatGPT-User to retrieve the actual content from selected URLs. The namespace blind spot This may be our most surprising finding. The trail started with classic reverse engineering. We decompiled the ChatGPT mobile app, dissected the web client source code, and sniffed network packets on both platforms. That gave us the names of internal tools and some calling conventions. Armed with these specifics, we were able to ask ChatGPT the right questions, and discovered the model answered without any restrictions. OpenAI has real safeguards around its system prompts. But the internal tool configuration layer has none. ChatGPT’s namespaces — the groups of internal tools the model can call during a conversation — are freely describable. As long as you avoid the words “system prompt,” the model will disclose tool schemas, operation lists, output channels, and namespace structures with perfect consistency. We published ready-to-use prompts that anyone can paste into ChatGPT to audit its internal environment. To verify that the model wasn’t hallucinating these descriptions, we ran a participatory study with dozens of users across separate sessions. Every participant got exactly the same tool names, parameter schemas, and operation lists. The model consistently and reliably describes its own tooling. The study also includes a reconstructed system prompt extracted progressively, along with several notable findings: Reddit is the only domain exempted from copyright word limits. There is a granular list of banned products. A “verbosity score” operates on a 1–10 scale. A full advertising policy paragraph governs ad display by subscription tier. Practical use: running your own crawlability audit The web.run syntax we documented isn’t just a technical curiosity. It works, and it opens a direct path for testing how ChatGPT interacts with your content. Here’s a concrete example. You can force ChatGPT to search your domain and read specific pages by pasting JSON commands directly into a conversation. First, trigger a targeted search on your site, then force it to fetch the first two results, then ask it to return the title, main topic, and key points from each page. "Search for this query, then open the first two results and summarize what you find on each page. Step 1: Search: { “search_query”: [ { “q”: “site:abondance.com seo” } ], “response_length”: “short” } Step 2: Open the first two results: { “open”: [ { “ref_id”: “turn0search0” }, { “ref_id”: “turn0search1” } ] } Step 3: Give me a structured recap of what you found on each URL. For each page: the title, the main topic, and 3–5 key points." What you get is a view of your content through ChatGPT’s eyes: what it can actually reach, what it extracts, and how it interprets your pages. If ChatGPT can’t access a page, returns garbled content, or completely misses your main messages, that’s a signal to act on. Same model family, different citations GPT-5.2, 5.3, and 5.4 share the same knowledge cutoff (August 2025) and belong to the same GPT-5 family. Yet the same prompt sent to each produces different fan-out queries, retrieves different sources, and surfaces different passages in the final response. Multiple layers of divergence come into play after pre-training: RLHF reward shaping, supervised fine-tuning data, system prompt configurations, and inference-time compute budgets. GPT-5.4 Pro explicitly gets more compute to “think harder,” and that alone can change which sources are cited. This is why we recommend testing model by model. A single prompt can produce radically different citations depending on whether the user is on GPT-5.3 Instant, 5.4 Thinking, or 5.4 Extended. Free-plan users may also be silently routed to a lighter model. Two types of AI visibility Our study introduces a framework that separates parametric visibility (what the model learns from training data with search disabled) from dynamic visibility (what it retrieves in real time with search enabled). Parametric visibility: E-E-A-T for LLMs. Parametric visibility is the E-E-A-T equivalent for large language models. It’s authority encoded across billions of training examples, shaped by press coverage, Wikipedia presence, other high-authority sites, and the overall training corpus. It’s stable and measurable through one-shot API audits. Dynamic visibility: shifting ground. Dynamic visibility is volatile. It’s model-dependent and requires continuous monitoring. It’s closer to traditional SEO, and can collapse overnight with a model update, as the Bigfoot Effect shows. The link between the two matters. The model formulates its web queries by targeting sources it already knows. A brand absent from parametric memory won’t even be considered as a search candidate. Being unknown to the model means being invisible before the search even starts. Knowledge cutoff updates are the “Google Dance” of LLMs. When the cutoff date changes, parametric rankings are redistributed in bulk. But this only happens roughly once a year, because retraining at that scale is extremely expensive. The strategic window for influencing what the model knows about your brand sits between two cutoff dates. Dan Petrovic’s (DEJAN) AI Brand Authority Index illustrates parametric measurement at scale. Our study complements it with a lighter, reproducible testing framework based on five prompts run multiple times for a one-shot audit. Dig deeper The full study — including reverse-engineered documentation, the honeypot experiment, DIY audit prompts, and the reconstructed system prompt — is available at think.resoneo.com/chatgpt/5.3-5.4/. Bottom line ChatGPT Search is no longer a black box. This study maps its internal architecture, from the web.run tool that powers every search to the fan-out logic that decides which domains are fetched and which are ignored. The 20% drop in cited domains after the switch to 5.3 shows how fast the citation landscape can shift with a single model update. But the deeper issue is structural: ChatGPT is concentrating citations on fewer websites and applying source selection logic shaped by training data, post-training fine-tuning, and system prompt rules that change from one model to the next. Tracking visibility in ChatGPT means understanding two distinct layers (parametric and dynamic), testing across multiple models, and monitoring a system whose internal tools are documentable but whose behavior can change overnight. The full study provides the data, methodology, and tools to get started. View the full article

Dozens of brands are using the 2026 FIFA World Cup as a chance to cash in on themed ads, products, and brand collaborations. But the home goods giant Lowe’s is doing something unique: debuting a 10-foot-tall inflatable of Lionel Messi for fans to put in their front yards. Lowe’s is running a series of activations for the world’s biggest soccer moment, all of which center on its limited-edition, $99 Messi inflatable, made in collaboration with Messi himself. The inflatable, which will start to pop up in a 20-foot version around several U.S. host cities in mid-May, will be available online to Lowe’s rewards members starting on May 18, followed by a limited release in select stores on May 20. According to Jen Wilson, Lowe’s chief marketing officer, the company is planning to release only about 5,200 inflatable Messis—and it expects them all to sell out. The reason for Wilson’s confidence is twofold: First, she says, while plenty of brands will be planning their own activations for the World Cup, not many others could even attempt a product in this niche. And, second, the move is backed up by company data that yard decor—especially personalized decor—is becoming more popular among consumers, even outside of the typical holiday windows. It’s a trend that, oddly enough, might just trace all the way back to a giant skeleton that stole the internet’s heart in 2020. What in the world is going on with yard decor? Over the past few years, I’ve been noticing a trend in my Chicago neighborhood. Outside the typical festive months of October through January, I’m seeing more and more holiday decorations left out in people’s yards and stylized for each new season. Oftentimes, that takes the form of a giant skeleton dressed up in a personalized outfit or performing some kind of goofy stunt. There’s a very real subculture to back this up, and it all stems from a giant Home Depot decoration. In 2020, Home Depot released a 12-foot-tall skeleton decoration that almost instantly went viral, earning the internet moniker “Skelly.” In the years since then, Skelly has become the only Halloween product that Home Depot brings back year after year, consistently selling out to its legion of fans. Skelly has amassed a cult-like following, and, in turn, inspired a small but committed group of decorators to keep their skeleton decor up year-round, giving them customized outfits and accessories for events like back to school and arranging them in silly poses like a staged flag football game. Skelly’s popularity seems to point to a broader shift in how Americans view their yard space. Wilson says that Lowe’s also saw consumers’ interest in out-of-the-box yard decor spike starting back in 2020—and the trend has only grown since then. “For us it was really this explosion of both all things mini and all things giant,” Wilson says. She believes Lowe’s was one of the first companies experimenting with products like mini buckets or mini toolboxes, which have become huge fan favorites. On the other end of the spectrum, like Home Depot, Lowe’s has begun investing in new giant animatronics, including its popular 10-foot Abominable Snowman, 8-foot Skelly-esque skeleton, and 12-foot-tall Immortal Nightwalker. Outside of the holidays, the brand has noticed and capitalized on year-round yard trends, like the “porch goose,” a TikTok-viral concept wherein customers buy a concrete goose and dress it up seasonally—just like some Home Depot fans with their beloved Skellies. “We do absolutely see a rising trend in outdoor decor and consumers either keeping outdoor decor up longer or participating in trends like the porch goose,” Wilson says. “It’s all really interconnected to expressing your own sense of style and culture and just being a part of something.” She attributes the rising consumer interest in these novelty products to something she describes as “similar to the lipstick effect,” or the idea that consumers will increase spending on small luxuries during moments of economic strain. “People still want indulgences, even if there’s a pressured economy,” Wilson says. “A larger-than-life item in their front yard is something that just makes them feel joyous, and that’s what people are looking for.” With the Messi inflatable, Lowe’s is betting that the same theory will apply to World Cup fans who are watching the games from their homes, but want a way to let their whole neighborhood know that they’re part of a larger moment. The Skelly of soccer Given Lowe’s “affordable indulgences” decor theory, one of the in-house design team’s key considerations when building the inflatable Messi was cost. “We wanted it to be under $100, particularly as people are paying attention to their wallets and obviously the rising costs of gas,” Wilson says. “When we know that the consumer is super focused on essentials, if they’re going to make this splurge, we want to make sure that it’s affordable. We typically look at engineering most of our gigantic items somewhere in the under-$300 range.” That price constraint helped Wilson’s team to determine the actual height of the inflatable, opting for 10 feet rather than 15 or 20 in order to conserve materials and lower costs. Then, to get every detail just right, the team went directly to Messi to determine how the inflatable should look: each aspect, from the length of his hair to his beard, tattoos, and the look of his arms and legs, was given Messi’s final approval. “We just wanted it to feel authentic and for him to be proud if he was driving down a neighborhood in Miami and saw himself outside of a home,” Wilson says. “He loved where we landed, and we’re thrilled.” View the full article

“Who are your enemies?” I was asked this interview question throughout my entire career. And I’d always come up blank. Every time. No enemies. And when I failed to produce an impressive enemy list, the reaction was always the same: How can you claim to be competent if you haven’t made powerful enemies? I came to understand this enemy thing was rooted in the male idea of power. That men tend to see winning and power like this: For me to win, you need to lose. I came to realize that this advice to be powerful enough to have enemies was basically an invitation to turn into an aggressive bully to advance my career. But here’s the catch. I was bullied as a kid. And it was awful. So, early on, I decided that I was never going to choose to be like the bullies who hurt me. And if that was not good for my career, so be it. I would find another way. I often wondered if I was limiting my career by being too nice. And worried if I was supposed to feel powerful? Am I supposed to act powerful even if I don’t feel powerful? Am I doing the job of a leader wrong because I don’t feel powerful? Even when I was in my biggest roles, where I had actual power at my fingertips — thousands of employees under my watch, millions of dollars of budget to manage, billions of dollars of revenue to keep growing — I never felt personally powerful. Mostly, I personally felt crushing responsibility. I felt insecure about the power thing for years. The VP Bully Then one day, what I needed to do about this idea of acting powerful like the men became very clear to me. I was at a client’s office on Long Island. Sitting in a small conference room were the VP of technology, who was a large, dominant type, and one of his direct reports, whom I’ll call Seth. The VP told me, “The reason we are having this problem is that Seth makes stupid mistakes. He’s not good at his job. No one listens to Seth. He screws everything up.” Seth looked small and mortified. I was cringing and heartbroken for him. I knew what it felt like to be bullied like this. “Little Patty,” who had been bullied herself, could feel her childhood insecurities and fears bubble up watching this VP berate Seth. I had worked with Seth on prior occasions. Seth knew a hundred times more than this VP. The problem was not Seth. This VP was a bully. But then a really weird, creepy thing happened a bit later, when the VP walked me out, and we ran into his boss in the lobby: this bully instantly became a cowering suck up to his boss. I was appalled. He needed to abuse Seth to feel powerful, but he was afraid to be powerful with his boss. Watching this scenario, a new thought started to brew: Wait a minute, if I am still the same insecure little kid on the inside, probably so is this jerk. And once I saw it, I couldn’t unsee it. Forever after leaving that lobby, whenever I see a big, scary man acting like a powerful bully, I see the hurt little boy, as plain as day. I want to reach over, gently squeeze his forearm, and say, “Aw, did somebody steal your ball? Did your father yell at you for crying about it? Poor thing.” A better way Seeing the big bullies as fragile little boys was my first step toward understanding that there was a better way to show up as a leader than “powerful.” And with this insight, when I got bullied at work, I could mostly just step aside and let the aggression roll by instead of being crushed by it. My mom had given me the key to keeping my self-esteem intact with bullies all those years ago. And I have used her advice for the entirety of my career and life: Bullies need to make you feel worse than they feel on the inside. It’s always about them. It’s never about you. Once I saw these men as their own little version of Kevin or Harold, struggling with their own insecurities, I was no longer worried that they were innately gifted with a kind of power that I didn’t have access to. It made me stop worrying once and for all about feeling or even acting powerful. It just didn’t matter. I chose to be a leader who was first and foremost kind and respectful to people. People are not productive when they are self-protecting. I focused on making people feel safe. My teams executed on our commitments. We grew the business. My organizations got more capable over time because I invested in and cared about the people. For me, real power is not personally owned. The aggressive, bullying version of personal power is just insecurity masquerading as strength. Sharing power with others so you can get big, amazing things done together is true power. That was the sort of power I chose to cultivate and the kind of leader I chose to be. Do aggressive bullies get ahead? Yes, of course they do. But I learned it’s not the only option. You can make a different choice. I made a different choice. I chose not to model the idea of power that was being shown to me by the men. You might say I chose to stay “too nice”. And you know what? It did not limit my career. If anything, it accelerated it. I was able to build a highly capable team of people who were productive and motivated because I chose to make them feel powerful. And the idea of being or acting powerful personally didn’t confuse me anymore. I had no interest in the win-lose version of power. I just let the men duke it out among themselves, and I created my own path forward that was true to my belief that kindness and strength can go hand in hand. Because making people feel respected and safe makes them wildly productive. And on the enemies thing, I just think, if I can win and you can win, why is that not better? View the full article

Over the past day or so, I am seeing a large spike in signs that there is Google search ranking movement and volatility. This is between the SEO community chatter spiking, as well as some of the tools showing some big volatility swings.View the full article

Google announced that GA4, Google Analytics, has a new AI Assistant traffic measurement. This allows you to track AI chatbot traffic, like from ChatGPT, Gemini, and Claude. This is through a new AI Assistant channel in your Default Channel Group reports.View the full article

The critical challenge is to build institutions that protect us from tech companies and the stateView the full article

Google is expanding its vehicle ads to support more vehicles in the United States. Specifically, All-Terrain Vehicles (ATVs), RV's Utility Task Vehicles (UTVs), and non-motorized trailers, such as travel trailers and campers are now supported in Google Vehicle Ads.View the full article

A new AI citation study sparked a familiar SEO debate: the difference between a ranking factor and a symptom of ranking success. The post Direct Traffic & Popularity – Correlation, Not Causation appeared first on Search Engine Journal. View the full article

Google sent out emails to some advertisers notifying them that after June 10, 2026 they will automatically link your Google Ads account with your YouTube channel. Linking them will give Google Ads data on your YouTube channel.View the full article

Google has released version 24.1 of the Google Ads API, this is a minor release with dozens of updates. This update includes features like mobile device platform segment, classic images in DemandGen, support for passkeys, revamped support of experiments, and support for the new data retention policy.View the full article

Disclosure: I’m the co-founder of Optmyzr. I’ll use one of our open-source skills as the example below, but the frameworks here apply to anything you install or build. If you’ve used Claude, ChatGPT, or Gemini for marketing work in the last six months, you’ve probably hit the same wall I have. The chat is great until you need the same thing done the same way every week. Then you’re back to copying a prompt template into a fresh window, hoping you didn’t forget a step, wondering why a tool this powerful still feels this manual. Skills are what bring that wall down. I’ve written before about skills as scalable systems for PPC and why agents are useless without access to your marketing data. This piece zooms out. What is a skill, actually? Where do you find them? How do they work across the three big AI platforms? But perhaps most importantly for agency owners: how do you take an existing skill and brand it as your own? What are ‘skills’ in AI A skill is a small bundle of files that teaches an AI assistant how to do one specific job well, every time. In Claude, a skill is literally a folder containing a SKILL.md file with instructions, alongside optional code scripts and reference files the AI can process. Install the folder once, and from then on, when a task matches the skill, the assistant loads the playbook and follows it. It’s the difference between telling a new hire “audit this account” and handing them your agency’s documented audit process. The output gets much more consistent. While the concept is universal, implementation varies by platform. Claude offers the most seamless experience, allowing you to install and use skills directly within the interface. ChatGPT makes similar capabilities available though generally only on paid Business or Enterprise plans. Gemini remains the most developer-focused, often requiring the Gemini CLI or specialized environments, which makes it less accessible for the average marketer. Because of its ease of use, I mostly use skills with Claude, and that’s where we’ll focus. Where to find prebuilt skills for PPC Most account managers prefer copy-pasting scripts over writing their own. They’ll also like grabbing skills someone else already built. But finding them can be tricky. There’s no single App Store for skills, and most of the good ones are on GitHub. For Claude, the Anthropic team ships official skills for working with things like PDFs, and Microsoft Office programs. Beyond that, you’ll find growing collections on GitHub from individual developers and software vendors. A lot of companies are publishing their own. Ours live at github.com/optmyzr-skills. Optmyzr’s free Google Ads audit skill. Screenshot by author of Github.com. May 2026 A practical rule I’ve landed on: a skill is only as trustworthy as the team that built it. A skill from a known software vendor with its methodology is different from a one-off prompt repackaged as a skill by someone you’ve never heard of. Ensuring skill usage is consistent across your organization This is where it gets interesting for agencies and in-house teams. On a solo plan, you install a skill in your own account and you’re the only one using it. Fine for a freelancer. Painful for a team, because everyone has to install everything separately, and versions drift the moment one person updates and another doesn’t. On Team and Enterprise plans, an admin can deploy skills across the whole organization. Claude has org-level skill management on Claude for Work and Enterprise. The practical benefit is that, with a five-person PPC team, you install a shared audit skill at the org level once, and every account manager gets the same version on day one. When you improve it, everyone gets the improvement automatically. No more “which version are you running” on team calls. How to install a skill in Claude. Screenshot by author of claude.ai. May 2026 I think of it like the moment ad scripts stopped living in each individual ad account and moved to Enhanced Scripts from Optmyzr, which lets advertisers deploy a generic script code to all accounts and shifts script versioning and settings into a centralized management system maintained by Optmyzr. Easy, maintainable, and scalable; all things that matter a lot to account teams who promise a standard of quality to their stakeholders. The hidden white-labeling engine: Why forkable skills are an agency’s best friend Here’s the part that should perk up agency owners. Most well-built skills are folders, and the open-source ones live on GitHub. Which means you can fork them, edit them, brand them, and use your modified version with your own clients. Let me walk you through an example for a 15-person agency. I find an open-source Google Ads audit skill (we’ll get to the one I’m thinking of in a second). The default report it generates has “Google Ads Audit” at the top in plain text. Useful, but generic and probably not shareable with the client. What I can do in about an hour: Clone or download the skill folder. Open SKILL.md and edit the report-generation instructions to swap in my agency’s name, reference my logo, and use my brand colors. Drop my logo into the skill folder so the assistant can use it when generating PDFs or HTML reports. Add or remove checks based on what I actually care about — if I run only ecommerce accounts, I can tell the skill to weight Performance Max, Merchant Center, and feed health more heavily. Repackage and install for my team. What comes out the other side is a branded, agency-specific audit tool that produces client-ready PDFs with my name on them. I didn’t have to build the underlying methodology. The original author already did that. I just added the last 10% that makes the output feel like mine. Scripts were powerful because you could tweak them. Skills have the same power. Agencies are no longer capped by what software vendors choose to white-label. If a skill is open source, you can white-label it yourself in an afternoon. A worked example: the Google Ads Audit skill Since I keep alluding to it: we recently released a free, open-source Google Ads audit skill at github.com/optmyzr-skills/google-ads-audit. Apache 2.0 license, no Optmyzr login required. A sample of the audit score the Optmyzr audit skill produces. Screenshot by author, May 2026. Briefly, what it does. It runs through 14 categories and roughly 42 best-practice checks: Account settings Conversion tracking Campaign structure Performance Max Budgets Bidding Targeting Audiences Keywords Quality Score Search terms RSAs Extensions Landing pages Industry benchmarks Competitor analysis It asks three calibration questions at the start (primary goal, target CPA or ROAS, account maturity) so the scoring matches the kind of account it’s looking at. If you don’t want to connect anything, there’s a four-paste flow: pull four CSVs from the Google Ads UI, paste them into Claude, and the skill runs the diagnostic. The output is a top-5 findings list with monthly dollar impact, an A/B/C grade with per-category breakdowns, a 7-day action plan, and a wasted-spend estimate. A sample of the next steps the Optmyzr audit skill suggests to address key shortcomings of a Google Ads account. Screenshot by author, May 2026. All the principles I described above apply to it. You can install it for free. You can deploy it across your agency. You can fork it, brand it, and have it generate client PDFs with your logo and your methodology framing. The Apache license explicitly allows that. If you want it to also pull live account data instead of CSVs, run multi-account portfolio rollups, and trigger automated remediation, that’s where Optmyzr’s MCP server comes in — and that’s the paid layer. But the audit logic itself is yours to use, modify, and brand. What to do with this Pick one repeatable workflow your team does manually right now. Audits, search term reviews, ad copy generation, weekly report drafting — anything that runs the same shape every time is a candidate. Find or build a skill for it. Then move it from individual installs to team deployment. That single change kills a surprising amount of version drift across a team. Brand at least one skill as your own, even if you never ship the branded version to clients. Going through the fork-and-modify process once changes how you think about what counts as “tooling” for your agency. It’s lighter than you’d expect. Skills are how a generic chatbot starts to behave like your team’s documented operating system. The agencies that get fluent with installing, deploying, and forking them over the next 12 months are going to operate noticeably differently than the ones still copying prompts into chat windows. The Optmyzr audit is one example. There will be hundreds. How to install the audit skill There are two paths; pick the one that fits your team: If you have someone technical on the team — an in-house dev, a power-user analyst, anyone comfortable with a GitHub URL — install it as a Claude plugin straight from the repo. One command. The skill stays in sync. When we ship a new check, tighten a benchmark, or add a category, your install picks it up automatically. For an agency running this across a team, this is the right path. Everyone runs the same version. If you’re a marketer who just wants the thing to work — no GitHub, no command line — download the zip from the repo’s releases page and upload it via Settings > Capabilities > Skills inside Claude Desktop or claude.ai. Up and running in under a minute. The tradeoff: it’s a snapshot. When we improve the skill, you’ll need to grab a new zip and re-upload. For a solo practitioner running monthly audits, that’s usually fine. Either way, the repo is at github.com/optmyzr-skills/google-ads-audit. Once it’s installed, type /audit in any Claude conversation, answer a few guiding questions, and then receive the audit. View the full article

Custom AI models are not just for the AI giants anymore. Because the 37-person startup Krea is releasing its first generative AI model as the design tools startup repositions itself as a full-fledged AI research lab. The move is significant for Krea, but it also seems to tease an almost inevitable moment in the rapidly evolving AI market, where smaller players in the industry can make more disruptive bets. On one hand, Krea can hardly call itself a bootstrapped startup anymore. It’s now raised $83 million through its Series B at a $500 million valuation. On the other, it’s tiny compared to the leading frontier model companies, which constantly raise more money to ensure they have an unlimited war chest to train the next best model: OpenAI and Anthropic, which have raised $180 billion and $72 billion, respectively. But to Krea’s co-founder, Diego Rodriguez, it’s invigorating to be small, nimble, and, by one significant measure, no less successful than any frontier model company as a core business. “Until there’s a winner—until OpenAI or someone is profitable—the Olympic Games are on,” he says with a mischievous smile. The evolution of Krea Krea launched in 2023 to be something like the Adobe of the AI age, a creative platform designed from scratch to allow you to not just generate media with AI, but to tune those outputs, with controls that feel more like a synthesizer than a drafting table. They were the first to offer real-time AI editing tools and the first to put APIs from other AI models into their own app (a practice that has now become standard). And they were quickly profitable. But over time, the team has recognized a distinct ceiling to their work: Krea can only be as open-ended as the models it sits upon. Image models of today are amazing at specific prompts that often go viral, but they can also feel like they are built on rails. Creative phrasings can still lead you down the same old paths, as models fail to reproduce what’s in your mind’s eye. “The models are trained not to fail and to always give you a good image,” says Krea’s co-founder, Victor Perez. “And I feel like that takes away a lot of the creative uses—breaking the barriers and letting people go off-road, letting [you] make ‘bad’ images, stuff that looks more artistic that a creative might appreciate more.” Indeed, image models are amazing when it comes to what these companies have been prioritizing: photorealism. But any designer reading this knows that when it comes to graphic design and illustration, you can hit the boundaries faster than you’d think. In a demo, Krea pulls up comparisons of the prompt “a cat riding a bicycle” between itself and Google’s Nano Banana. In Krea’s case, the first outputs are funky and varied, with some exhibiting a hand-drawn feel. In Google’s, no matter how you adjust the prompt, you get a similar coloring-book-looking image presented in the same way. It’s the difference between eating at McDonald’s or a Michelin burger joint. One will always aim to please, while the other may polarize. “I think that the kind of stuff that we are interested in is more niche,” says Perez. “It’s a much smaller market, but we’re fine with it.” Spending 15 minutes prompting Krea’s new image model K2 on my own, and I’m impressed by its breadth. It generates surreal photorealistic scenes, but also grainy VHS-style filtered images and a variety of illustrative techniques (word marks, manga, anime, hand sketching, and sharpie cartoons) well. The examples I saw from Krea were also impressive—and wildly so given the gulf in resources between Krea and the giants. Perez attributes this success to his team’s own taste. They’ve spent the last seven months building their own data set (no, they aren’t disclosing the sources), labeling it by hand, and creating their own unique workflows to train their own generative AI system. As Perez explains, most big models start the same in development to build a functional neural net, but mid- and post-training steps in particular are what give the model a point of view. I’ve heard from people in the industry that there are only about 200 true post-training experts in the world, which is why the market is so competitive. “That’s when the artistic direction on the model takes place,” says Perez. “At the end of the day, building a model is almost like crafting a sculpture.” That final layer of training, where a model develops its visual or verbal voice, is where taste comes in. Making the AI do one thing better can often make it do another thing worse, and balancing those priorities is particularly tricky when trying to build a model that makes cool, personally expressive stuff. “This is like the nemesis of an AI researcher, because what researchers are really good at optimizing for [is] metrics,” says Perez. “But what is this metric that we are optimizing for? Like, it’s something so subjective.” The user interface K2, Krea’s new model, seems impressive on its own. But what makes it so attractive is how Krea will let you use it. On the baseline, Krea promises that just describing what you want will get you better results with K2 than its competitors. Then Krea’s user interface lets you really get your hands dirty in tuning the output. You can drag one or multiple images you want into the prompt bar, to use that to influence the style it generates. Then you can drag a slider up or down on those images, to signal how much you want them to influence the visual style. You can even build a mood board to inform the aesthetic that you’re after. (After generating some images, Krea will proactively produce a sort of personalized Pinterest board with more images it thinks you’ll like.) Because this system is built for creatives, Krea is also being careful with how it frames up IP. As you ostensibly train your own model inside Krea, you can remove that from Krea’s own model training. And all IP generated is your own. So if you are an oil painter who has a very particular style that you want to use within gen AI media, you can upload your work to reproduce it without worrying that Krea is about to sell that as a filter to someone else. Longer term, Krea is considering if there are ways to credit artists whose IP measurably influences a piece of media, and they’re experimenting with using AI to do just that to create a more sustainable royalties system. Rodriguez admits some confusion as to why, in an industry dominated by OpenAI, Anthropic, and Google, smaller AI companies aren’t banding together in order to build bigger ideas and share the wealth. Originally, Krea tried partnering with a model company that refused to offer even a small split of revenue, which led them to develop the technology completely in-house. But now, I can’t help but notice how much Krea’s ambitions have grown. Perez declares that this launch product, K2, is “conservative.” The GPU cluster Krea is using for a year, over which time it will have trained K2 and two future Krea models, will cost the company $20 million. Krea couldn’t afford to faceplant with an experimental approach that might not work. However, with a success under their belts, they feel more confident to take more risks and challenge training norms. “We just wanted to make it work,” says Perez. “It worked way better than we expected, but this was an extremely risky bet. We’d never trained a model before. We didn’t know how hard it would be. And it was it was fucking hard, but at the end of the day we figured it out. And now we know so many things—because there’s so many things about training a model that you can only learn through training a model.” View the full article

If you were to travel back in time to 1996 with a 2TB thumb drive, you’d be able to fit the entire World Wide Web on it. Of course, that kind of storage didn’t exist in the ’90s, so it’s never been that simple for the Internet Archive. The nonprofit site, which launched three decades ago this year, went from making copies of the web on tape drives to storing more than 1 trillion pages worth of Internet history at data centers around the world. Using its Wayback Machine, anyone can look back to what a web page used to look like, which means you can browse through old GeoCities websites, view Google’s original Code of Conduct (back when it still said “Don’t Be Evil”), or read the EPA’s climate change indicators before the The President administration scrubbed them. All that’s on top of the Archive’s vast collection of other digital resources, from live concert tapings and public domain e-books to troves of forgotten DOS games. Roughly 2 million people access the site’s resources every day. “We want it all,” says Brewster Kahle, the Internet Archive’s founder and chairman. “We want all the public works of human beings. So if we don’t have it, we want it.” But while the Internet Archive hasn’t fundamentally changed over the years, the Internet itself is transforming in ways that jeopardize the nonprofit’s mission. Web publishers have started blocking the Wayback Machine out of fear that AI companies are scraping the material. A legal battle with book publishers ended with the Archive paying a settlement and removing more than 500,000 books from its collection. Meanwhile, the cost of storing humanity’s digital footprint keeps going up, as demand from AI data centers drives up storage and memory prices. All of which makes Kahle wistful for how things used to be for the Internet Archive, before book publishers, tech giants, and the legal system got in the way. “We have to still try to make a library work, even though it’s a difficult, difficult time for libraries,” he says. The Internet Archive isn’t just a way to access old web pages, important as that may be. It’s also a repository for information and culture that anyone can access, download, and do what they please with. In a world where digital content is increasingly licensed rather than owned, that in itself seems like something worth preserving. How it started Kahle had been dreaming of something like the Internet Archive long before it became feasible. In the early 1980s, he studied AI at MIT and became a lead engineer on supercomputers at Thinking Machines. The modern internet wasn’t born yet, but he recalls imagining that these supercomputers would someday make reference materials readily available to anyone. “For me, back in 1980, the idea was to try to build this thing that we’d long since promised by then, which was the Library of Congress on your desk,” he says. The real epiphany, though, came in 1995 while Kahle was visiting the offices of AltaVista, one of the first Internet search engines. While early work on the internet had focused on decentralized protocols, AltaVista had built something useful by providing a hub of all the Internet’s knowledge. Kahle realized the same crawling technology could help make full copies of web pages for archival purposes, which AltaVista wasn’t interested in doing. “I thought that the key was making sure that the works of humankind would be preserved, so we went off to collect it,” he says. Kahle kicked in some of his own money to start the Internet Archive—he’d sold an early web publishing system called WAIS to AOL for shares worth $15 million, after spinning it off from his work at Thinking Machines—and got some help from outside backers. But the real heavy lifting came from Alexa Internet, the for-profit traffic analysis company that he founded at the same time as the Internet Archive. For every web page that Alexa crawled, it donated a copy to the Internet Archive, and Kahle made sure that arrangement endured even after Amazon acquired Alexa for $250 million in 1999. Amazon quietly contributed to the Wayback Machine for more than 20 years, until it shut down Alexa Internet in 2021. (The Alexa name, which was based on the Library of Alexandria, lives on as the name of Amazon’s virtual assistant.) “My hat is off to Amazon,” Kahle says. “They could have figured out how to get out of that contract, but they didn’t. So it really gave the Internet Archive, when it was a very young nonprofit, a content set.” Running the Archive The Wayback Machine was rudimentary at first, relying on simple automations to capture the code behind each webpage, preserving what they said and looked like at that moment. Over time, it’s become increasingly sophisticated, with new crawling engines aimed at capturing the growing complexities of the modern web. These days, the Wayback Machine takes snapshots of roughly 1 billion URLs per day. It maintains copies of more than 1 trillion web pages, and stores 100 terabytes of new data per day in the process. Still, Kahle says the Wayback Machine represents only about 60% of the Internet Archive’s data. The rest comes from its vast digital collections, including radio shows, podcasts, defunct mobile apps, DOS games, CD-ROM software, publicly available scientific research, scans of vintage magazines, classic TV shows, past cable news broadcasts, documents scanned from microfiche, and more. Both sides of the Internet Archive share the same computing resources. Despite the scale at which it operates, running the Archive is a surprisingly human endeavor. While the site has tens of thousands of automated processes for archiving the web, its resources are ultimately limited, and it often needs to set priorities, says Mark Graham, the Wayback Machine’s director. “Part of what I do every day is pay attention to this process, through conversations, through examining what we’re archiving and maybe what we’re not archiving,” Graham says. Graham recalls a recent example in which the State Department revealed plans to delete its posts on X from before Donald The President returned to office. He quickly spun up a project with his team and ultimately saved more than 2 million posts, hundreds of thousands of which have since vanished from their original URLs. Graham’s team has also made emergency copies of online publications whose shutdown is imminent, as he did recently with a prominent gaming site (which he declined to identify). “We’re notified almost every day about certain web properties that are going to be shut down,” Graham says. “Often we’ll get weeks or months of advance notice, but sometimes we don’t.” The Internet Archive doesn’t undertake all the work on its own. The group partners with more than 1,400 other groups, including libraries, universities, and museums that help decide what’s worth saving at any given time, and it operates a paid service called Archive-It for groups that want to maintain their own digital collections. Individual users can also archive pages manually through a web form or browser extension, and can even upload files for the Internet Archive’s digital collections. “It’s a healthy mixture of different methodologies, different motivations, different agency,” Graham says. Threats to the archive For most of its existence, the Internet Archive hummed along without much conflict. That’s started to change over the past few years. For the Wayback Machine, the web itself has become harder to archive. The Internet Archive doesn’t save paywalled articles, so it’s missing large swaths of content from major publishers. “It’s gotten a lot harder to do a good job of archiving the public web, because more and more of the web is not public,” Graham says. Some of those publishers have also started blocking the Internet Archive to prevent AI companies from scraping their content. Nieman Lab reported in January that 241 news sites explicitly block at least one of the Internet Archive’s crawling bots, most owned by the newspaper conglomerate USA Today Co. The French newspaper Le Monde has blocked the site as well, while The Guardian has filtered its articles from the main Wayback Machine interface. Reddit also began blocking the Internet Archive last year. Graham says the Internet Archive employs a variety of tactics to turn away AI scrapers, but acknowledges that this requires “nearly constant care and feeding.” Jack Cushman, director of the Harvard Library Innovation Lab, says publishers may be largely indifferent to the work of archivists, at least compared with the more immediate threat of AI repurposing content or putting a strain on their servers. (Cushman’s lab has developed its own archiving tool, called Perma.cc, that it offers to individuals and institutions.) “The upshot is that the doors are slamming shut, incidentally keeping us out, when they don’t really care about us in the first place,” Cushman says. Meanwhile, AI is posing a threat in another way, in that demand from AI data centers is driving up the cost of storage. Kahle says the Internet Archive’s hard drive costs have already tripled to quadrupled as result. “We’re going to have to start becoming really clever about how to go and continue to archive,” he says. And as the cost of storage is going up, a growing proportion of what people consume online involves video on sites like YouTube and TikTok, taking up more space than static images and text. That means the Internet Archive must become even more selective about what it saves. Its YouTube collection is only in the millions of pages, versus more than a trillion web pages overall. “There’s other cases where there is just so much material on a given platform or service that we don’t have the capacity,” Graham says. Outside the realm of archiving the web, the Internet Archive’s digital collections have become a source of legal trouble. Book publishers sued the group in 2020, after it started lending out digital scans of physical books as a response to the COVID-19 pandemic. That resulted in an undisclosed settlement and the removal of 500,000 books from the Internet Archive’s collection. The group also settled a separate record label lawsuit over its collection of digitized 78 rpm records, though those remain available. Cushman says that those lawsuits have drawn attention to the well-intentioned risks that archivists take with copyrighted material. While the Internet Archive has typically avoided things that might upset copyright holders, that’s started to change in recent years. “They’ve moved into some things—especially with the pandemic—that really did anger some people with deep pockets, and great lawyers, and so on,” he says. “It makes the edifice a bit tippier in a way that I think that no one would have wanted.” Kahle and the Internet Archive see those lawsuits as a major detriment to its mission, one that further moves all content consumption to a model of licensing and surveillance, rather than ownership. “The United States has just kind of descended into just lawsuits, where in the ’90s, the United States was interested in innovation, and having a game with many winners,” Kahle says. The Internet Archive remains an indispensable resource, Cushman says, one that’s regarded among archivists as something of a benevolent monolith. There’s a playfulness in how it operates—for instance, in offering a playable collection of LCD gaming handhelds—that no one else is doing. But its challenges also make him wish there were more organizations trying to do similar things. “It’s different from anything else that we have,” Cushman says. “So I think we look at it with a mix of gratitude, where we’re fortunate that it happened, and then apprehension because there’s only one of it.” Looking ahead Kahle built his life’s work around digitizing the world’s knowledge and even using AI to make it more accessible. Now that future is finally materializing, but in a way that is, ironically, concentrated around a handful of well-funded tech companies, media conglomerates, and publishing giants. As a young engineer, that possibility was never on his radar. “I didn’t predict the monopolies,” he said. Kahle still sees AI as an opportunity to sort through the Internet Archive’s vast stores of data. Researchers are already using it, for instance, to do things like interpret key talking points on Russian newscasts, and the Internet Archive has been leaning on AI to help digitize and translate more content. But those opportunities, he says, are increasingly happening outside the United States, where there’s more legal certainty around what libraries can collect and digitize. The European Commission, for instance, is pursuing the concept of AI for the public good, promoting tools that tackle specific challenges like climate change and health care. The Internet Archive Europe, a separate group on which Kahle is a board member, has been backing a open-source tool called ClimateGPT that applies large language models to climate research. “There could be hundreds of innovative organizations going and conquering all sorts of niches, if they had the same kinds of policies in the United States that we had in the 1990s when we let search engines happen here,” Kahle says. Still, Kahle says he’s not discouraged, because fundamentally people want their works to be read and preserved. They also want good information that’s easily accessible, which is why the Internet Archive is being used now more than ever. And while the Internet Archive was born from the idea of centralizing the world’s knowledge, lately it’s been sponsoring conferences on ways to decentralize the web again. It’s early days, but he’s hopeful that this will lead to new business models that recapture what once seemed possible 30 years ago. “Let’s build systems that support communities,” Kahle says. “Let’s make tools for participation. Let’s build democracy’s library out of all the works that can and should be shared, so we’re all building on a common commons of information.” View the full article