The Mad Men era of SEO: Why AI is shifting search to persuasion

For most people, “Mad Men” means the TV show. But the phrase points to something more specific: Madison Avenue in the 1950s and ‘60s, when agencies grew brands through persuasion, positioning, and earned trust in a world of scarce media channels and powerful gatekeepers. If you wanted attention, you bought your way in, then made your product the obvious choice.

When the internet arrived and Google made the chaos navigable, an entire industry was built on getting brands found. Search and SEO became one of the most commercially valuable disciplines in marketing.

That model isn’t disappearing. But something new is taking shape on top of it — and most of the industry is still using the wrong language to describe what’s happening.

AI is exposing everything SEO has neglected. Brands that win recommendations from AI systems won’t do so by publishing more content. They’ll win through positioning, persuasion, and corroborated proof.

In other words, they’ll win the way Madison Avenue always did.

SEO was never really about content

One of the strangest things about the current industry conversation is how many people talk as if the job of SEO is to create content. It isn’t. Not for most businesses.

If you’re a publisher, content is the product. Traffic is the commercial engine. But for most brands, content never did what people thought.

Early on, people wrote content for customers, and it worked. Then it changed. Content became a keyword vehicle. “Get people to our site” replaced good marketing comms.

Traffic became a proxy for exposure. It worked because search rewarded retrieval: type a query, get a page, get a click. All you needed to sell that model was the belief that any traffic was good traffic. That traffic somehow led to revenue that your agency could keep delivering.

That model is now under serious pressure. 

Google and ChatGPT are increasingly taking the click. Every serious large language model is trying to satisfy informational intent before the user reaches the source. They aren’t trying to be better search engines. They’re trying to make search engines unnecessary — and that’s the entire point.

There’s too much information on the web. People don’t want to open 10 tabs and read five near-identical blog posts to find a basic answer. They want the answer. The AI systems exist precisely to give it to them.

So if informational retrieval gets absorbed into the interface, what remains? Marketing. That’s the part many SEOs are still not fully grappling with.

Dig deeper: The three AI research modes redefining search – and why brand wins

From place to preference

The cleanest way to understand this shift is through the “4 Ps” of marketing: product, price, place, and promotion.

Traditional SEO has been, almost entirely, a place discipline. It’s been about getting your products, services, or information onto the digital shelf when people go looking.

Keyword rankings are shelf position. Paid search is just a more expensive version of the same principle. In commercial search, you pay for premium placement in a digital aisle.

That still matters enormously.

Buyer-intent search remains valuable. Google hasn’t solved its commercial transition to a fully AI-led interface, and won’t overnight. Search is too important to Google’s revenue to disappear fast. But another layer is emerging above it, and this is the layer that most agencies aren’t yet equipped to compete on.

As AI systems become the first interaction point for more users, the game shifts from being present to being preferred.

Users don’t just search. They ask. They describe a problem. They want the best CRM for a mid-market SaaS company, the best estate agent in their area, the best sandwich shop near the office. And the system responds with recommendations.

If classic SEO was about rankings, the next phase is about recommendations. If classic SEO was about digital placement, the next phase is about shaping preference. And recommendation, in practice, is advertising.

Not a display banner. Not a 30-second TV spot. But advertising in the oldest and most commercially powerful sense: influencing the choice someone makes before they’ve even consciously made it.

An AI-generated recommendation is an invisible ad unit. It doesn’t bill by impression.

Why AI recommendations hit differently

When an LLM recommends a brand, it can’t know with certainty what will work best. So it infers. It weighs signals: past success, prominence, reviews, case studies, corroborating sources, and repeated associations between a brand and a specific type of problem.

Humans do something almost identical. 

Where performance is clearly bounded, we can identify a winner. We know who won the Oscar. We know which film topped the box office.

But when performance isn’t obvious in advance, we rely on proxies. We ask friends, read reviews, and scan for authority. We use familiarity, logic, and social proof to estimate what is likely to be right.

That’s exactly the territory AI recommendation is now entering — the consideration set problem. If I ask an LLM to find me a reliable accountant for a small business, I’m not asking it to retrieve a blog post. I’m asking it to build me a shortlist. 

Unlike traditional search, the recommendation layer is invisible to brands unless they test for it actively. You don’t see the prompt or the source chain. You don’t even know why one brand made the cut and another didn’t.

But the commercial effect is real, possibly stronger than anything traditional search produced. If you’re in the recommendation set, you’re in the running. If you’re absent, you’ve lost the sale before the conversation started.

Dig deeper: Rand Fishkin proved AI recommendations are inconsistent – here’s why and how to fix it

---

Your website is now an argument for preference

The first practical consequence: your website can no longer function like a polite digital brochure. Despite being optimized for search, many commercial web pages simply:

• Introduce the company.
• Gesture vaguely at services.
• Bury differentiation under generic corporate language.
• Treat the page as an endpoint for a ranking rather than a persuasive asset.

Still, they’re weak where it matters most: actual selling.

In the Mad Men era of SEO, your landing pages and service pages need to function like sales pages, not in a cheesy direct-response way, but in the strategic sense that they must clearly answer four things:

• Who is this for?
• What problem does it solve?
• Why is it different?
• Why choose it over the alternatives?

This comes down to positioning, which is key to GEO. If seven brands do broadly the same thing, the model needs distinctions. It needs enough clarity to say: this brand is best for X kind of buyer with Y kind of problem because it does Z better than everyone else.

Your website copy must surface real performance attributes: the specific things you genuinely do better or more distinctively than competitors. Your pages must become machine-readable arguments for preference.

Copywriting is back

Actual commercial copywriting — not fluffy brand storytelling or word count for its own sake — identifies a target customer, sharpens the problem, articulates the value, and makes the offer easy to recommend.

Good copy isn’t optional.

Take a local sandwich shop. The old SEO conversation runs to “best sandwich near me,” local pack, and review acquisition. It’s useful, but limited. 

The GEO version starts with the shop’s actual performance attributes. 

• Is it the speed? 
• The handmade bread? 
• The office catering? 
• The locally sourced produce?

Those claims must be clear on the website first. Then they need corroboration everywhere else:

• Reviews that mention the sourdough specifically.
• A local food blogger’s write-up.
• Inclusion in “best lunch spots” roundups.

They’re specific, repeated, retrievable evidence of why this shop is the right recommendation for a particular type of customer.

Scale that logic to a B2B software company, and the principle holds. Pages that clearly explain who the product is for, which problems it solves, and why it outperforms rivals. Then build mentions, customer reviews, and gain trade-press coverage — the body of evidence to support recommending you to buyers — and let the AI find it.

That’s pretty much GEO in a nutshell.

Keywords don’t disappear, but they lose their throne

Keywords are a human workaround. Approximations of intent, built for a retrieval system that needed exact string matching. LLMs process fuller context, layered needs, and comparative requirements. They move from keyword matching toward problem understanding.

Keyword research still matters for classic search, paid search, and buyer-intent pages. But the center of gravity shifts.

Instead of asking only “what terms should we rank for?”, the better question is: what attributes make us the right recommendation for the buyer we actually want, and what evidence exists across the web to support that claim?

The future of SEO is starting to look like the old agency model, as the work is increasingly promotional. Once your website clearly expresses your positioning, the challenge becomes promoting that position across the wider web through credible, repeated, relevant signals.

• Digital PR. 
• Traditional PR. 
• Expert commentary. 
• Case studies. 
• Reviews. 
• Listicles.
• Awards. 
• Trade press.
• Brand mentions. 
• Conference speaking. 
• Events. 
• Creator coverage. 
• Product comparisons. 
• Original data studies that other people actually cite. 

These are the things you go after, create, and encourage. Sadly, many “AI visibility” conversations flatten this into nonsense.

The goal isn’t merely to have content cited by AI. It’s to gather enough market evidence that AI systems repeatedly encounter your brand in the right contexts, with the right associations.

The work stops being optimization and becomes maximization: building the largest possible volume of persuasive, corroborated, retrievable evidence that your brand is a sensible recommendation for a specific kind of buyer.

That’s a fundamentally different model from anything the SEO industry has been selling. It’s promotional and strategic brand marketing.

Dig deeper: How to design content that AI systems prefer and promote

Where SEO still fits

SEOs need to grow up. There’s still significant value in buyer-intent search, technical site architecture, entity clarity, internal linking, and structured data. SEOs are well placed to monitor recommendation environments, test prompts, and identify where visibility is being won or lost.

But the identity crisis is real. Many agencies were built for a world of rankings, informational blogs, and monthly traffic graphs. They aren’t equipped to lead a world defined by positioning, copy, PR, brand evidence, and recommendation science.

Tracking brand citations inside AI outputs isn’t a complete strategy. It’s a temporary metric. 

The new agency model

Winning agencies look like hybrid commercial strategy firms: part SEO, part copywriting, part PR, part brand strategy, part technical infrastructure. They know how to protect buyer-intent search revenue today while building the fame, clarity, and corroborated authority that earns recommendation tomorrow.

This is the Mad Men model of SEO. Persuasion, positioning, and clear claims backed by public proof matter again. And the job is to become recommended by AI.

View the full article