Search, answer, and assistive engine optimization: A 3-part approach

SEO is all about optimizing for, well, search.

In 2018, I defined SEO as:

• “The art and science of persuading search engines such as Google, Bing, and Yahoo, to recommend your content to their users as the best solution to their problem.”

In 2025 (and beyond), we can define search, answer, and assistive engine optimization as 

• “The art and science of persuading recommendation engines such as Google, Bing, Yahoo, ChatGPT, Perplexity, Siri, Alexa, and Copilot to recommend your solution to their users as the best in the market.”

The aim is the same – get the conversion.

The difference?

• We have more engines.
• The recommendation is often further down the funnel.
• We have two additional foundational technologies to optimize for. 

3 technologies powering recommendation engines

At the heart of the future of search and research online lie three foundational technologies: 

• Large language model chatbots.
• Search engines.
• Knowledge graphs. 

Every search, answer, and assistive engine uses a unique combination of these three technologies, each blending them differently to deliver its own “flavor” of recommendations.

Large language model chatbots

The key functionality that LLM chatbots bring to the table is their ability to engage in conversation with people. 

This means they can directly answer questions, make suggestions, and actively guide the user toward the “best” solution to their problem.

Standalone, their weakness is that:

• They hallucinate (invent facts). 
• They are pretrained on a limited amount of data. 
• The information is never up to date (think: football scores and flight updates). 

The next two technologies are designed to solve those issues.

Dig deeper: Decoding LLMs – How to be visible in generative AI search results

Knowledge graphs

Knowledge graphs are huge machine-readable encyclopedias full of facts about entities (people, companies, films, topics, concepts, etc.). 

For example, Google’s Knowledge Graph is currently at least 10,000 times bigger than Wikipedia. 

Knowledge graphs are great for structuring information and fact-checking.

When you add a knowledge graph and a search engine to a chatbot, you solve the first problem: hallucinations. 

Dig deeper: When and how to use knowledge graphs and entities for SEO

Search engines

When you add a search engine that the chatbot can summarize, you expand its information source beyond its training data and ensure up-to-date responses.

Perplexity does this very well.

And, because the LLM chatbot can summarize the search results for the user, the need to rank first is no longer the key to success. 

Relevancy becomes more important since the engines will cite the most relevant reference from the top results for each piece of information it provides in its summary. 

Important: Here, even more than in traditional search, ranking applies across all types of results – not just blue links, but also news, videos, maps, books, shopping, images, and more.

With a hybrid search / knowledge / LLM chatbot result, the key is to hit one or more of these KPI:

• Be in the top 10-20 results.
• Have a presence in the knowledge graph.
• Be most relevant to the intent of the user.

Complementary strengths of the three technologies

LLM chatbots provide the ability to converse with the user.

Knowledge graphs provide fact-checking and topical context.

Search adds breadth and freshness to the information for the conversation. 

Historically, search engines came first, and Google continues to dominate, so for now, the “flavor” most people use is still search-first. 

From a user experience perspective, however, the LLM chatbot seems to be the more natural starting point. 

People are gradually moving toward that, and in a few years, conversational research will almost certainly take over from search. 

The key point to remember is that all these AI search, answer, assistive, and recommendation engines function fundamentally the same way, so the same strategy works for them all.

Dig deeper: 6 easy ways to adapt your SEO strategy for stronger AI visibility

---

How do different recommendation engines use the three technologies?

Note: The percentages I provide below are my rule-of-thumb guesstimates for the influence of each as of March. 

Don’t take these numbers literally; they are simply illustrative based on my experience and a bird’s-eye evaluation of the data we collect.

Google Search

With Google Search, the kickoff is search, with content from their knowledge graph in the form of:

• Knowledge panels.
• Entity lists.
• Topical filter pills.
• Other knowledge SERP features.
• Plus, LLM summaries (AI Overviews) for some queries.

Approximate blend: 

• 15% LLM.
• 25% knowledge graph.
• 60% search.

Bing Search

This is also primarily search with content from their knowledge graph in the form of: 

• Knowledge panels.
• Topical filter pills.
• A heavier integration of LLM than Google with generative AI content integrated directly into the SERP.
• Plus, a direct option of using their LLM chatbot, Copilot.

Approximate blend:

• 30% LLM.
• 15% knowledge graph.
• 55% search.

ChatGPT

ChatGPT is an LLM-driven chatbot, supplemented with search results, and limited use of a knowledge graph, if any.

Approximate blend:

• 65% LLM.
• 0% knowledge graph.
• 35% search.

Perplexity

Perplexity is fundamentally an engine that provides LLM-driven summaries of search results 

Approximate blend: 

• 50% LLM. 
• 0% knowledge graph.
• 50% search.

Google ‘Learn About’

This experimental feature is a clear indicator of where recommendation engines are headed. 

It’s the engine that is the closest to what I would imagine is the “ideal mix” of the three technologies, providing:

• LLM-driven summaries of search results. 
• Some fact-checking using their knowledge graph.
• Directly embedded search results.
• LLM and search-driven follow-up questions. 
• Contextual topical filter pills on the left-hand side (using both LLM and knowledge graph).

It’s designed to be a multimodal, contextually smart learning environment that guides the user toward the best solution (and from a marketing perspective, down the funnel to the perfect click).

Approximate blend: 

• 40% LLM. 
• 20% knowledge graph. 
• 40% search.

LLM chatbots, search engines, and knowledge graphs share the same data source

Optimizing for all three technologies may seem impossible at first. 

However, LLM chatbots, search engines, and knowledge graphs all get most of their information from one source: the web. 

LLM chatbots are pretrained on data collected from the web. 

Search results are generated in real time by information collected from the web.

Knowledge graphs are populated with facts extracted from the web. 

• The effectiveness of LLMs in generating accurate and informative responses.
• The relevance of search engine results.
• The comprehensiveness of knowledge graphs.

All three are intrinsically linked to the quality and structure of the information on the web.

That means optimizing your digital footprint is the key to optimizing for all search, answer, assistive, and recommendation engines, whatever the blend of the three foundational technologies now and in the future. 

You must focus on providing high-quality, accurate, and well-structured content across your entire digital ecosystem.

Hyper-optimize your small corner of the web.

By doing that, you will improve the probability you’ll: 

• Be included in LLM conversations.
• Be accurately and confidently understood by knowledge graphs.
• Rank in search.

What you can do to optimize for every flavor of recommendation engine

Since they all use the web as their data source, your ability to influence them stems from managing your digital footprint across the web.

SEO traditionally focuses primarily on ranking the website in search results.

Ranking the website is still valuable, but its primary aim is to provide a hub of clear and detailed information about the entity (company, person, book, film, etc.) that each of the three technologies can use as a reference.

Optimizing your entire webwide digital footprint and using your website to “join the dots” is the winning modern SEO strategy. 

Practical strategies for advanced SEOs

Entity optimization (Understandability)

The aim here is to ensure that the algorithms behind the LLM, search, and knowledge graphs can represent who you are and what you do.

For this, you need to create a clear and consistent set of facts across your entire digital footprint that is perfectly connected. 

You must implement the “hub, spoke, and wheel” model. 

• The About page on the website is the entity home hub where you state the facts (who you are, what you offer, and who you serve). 
• The external digital footprint is the wheel (that needs to consistently corroborate what you say on the entity home page). 
• You need to link from the hub to the different corroborative resources and back from the corroborative resources to the entity home, where possible. These are the spokes.

With that in place, the bots following links will go from the entity home to each corroborative source and back, consistently seeing the same information, and by pure repetition will understand.

E-E-A-T / N-E-E-A-T-T (Credibility)

You need to start by expressing your credibility (through notability, expertise, experience, authoritativeness, trustworthiness, and transparency) clearly on the entity home website (the hub of your digital presence) and across your entire digital footprint. 

Take the time to improve how you present your existing credibility signals, such as:

• Awards.
• Publications.
• Certifications.
• Reviews.
• Qualifications.
• Relationships with market leaders. 

Then, ensure this information is accessible for bots to find. This includes: 

• Your website.
• Second-party (controlled and semi-controlled) sites, such as:
  • Social media accounts. 
  • Crunchbase.
  • Etc. 

Amplify these signals further by getting them mentioned on relevant third-party sites like news outlets or industry associations.

Now, you can start to build additional credibility signals:

• Get more reviews from clients.
• Write a book, publish academic papers.
• Take a certification.
• Join an industry organization.
• Build relationships with market leaders. 

Communicate those on first, second, and third-party websites just as you did with your existing credibility signals.

Content (Deliverability)

Deliverability is all about your content strategy. 

Long gone are the days when text-only pages were enough. 

Today, you need multimedia – images, sound, video, and text. 

Make sure you help the bots (optimize) by adding textual clues for them, such as:

• Alt tags.
• Transcripts.
• Captions to your images, videos, and audio.

Traditional SEO focuses on publishing content on the brand’s website. 

That approach remains valid but needs to be expanded to second- and third-party sites. 

Off-site content is incredibly powerful and often overlooked. 

And don’t ignore user-generated content:

• Client reviews.
• Videos about your company or products.
• Articles.
• Social media posts. 

Be active and actively encourage your clients to actively talk about your brand and create content about it.

Preparing your SEO / AEO strategy for the future

The “secret sauce” is simple: no matter how different recommendation engines may look on the surface – whether it’s Google, ChatGPT, Perplexity, or Bing Copilot – they all rely on the same core ingredients. 

They use the web as their primary source of information for large language models, search engines, and knowledge graphs, blending these technologies to deliver solutions and answers.

By taking control of your digital footprint, you shape the exact information these systems use. 

You’re not just hoping to be found; you’re ensuring that the content they find is comprehensive, relevant, accurate, consistent, and aligned with your brand. 

That’s how you safeguard your SEO and AEO strategy, becoming the go-to recommendation in both search and AI.

View the full article