Following my recent webinar, I wanted to share the key ideas from a wide-ranging discussion with Mike King, founder and CEO at iPullRank.
Mike has spent years studying how the systems beneath search actually work. His view of AI, retrieval, and search mechanics is grounded in research and engineering, which makes him one of the clearest voices in the field right now.
Our conversation explored how AI tools reshape search behavior, why query fan-out changes everything, what GEO means as a discipline, and how the next wave of agents may treat content far differently than users do today.
This recap pulls those threads together.
Key resources to expand on this topic:
-
GEO A/B testing for ecommerce SEO: prove uplift from AI search by targeting fan-out queries and measuring the results compared to a control group.
-
Platform-backed GEO testing for retailers: run controlled experiments across PLPs and PDPs, track AI referrals vs blue-link clicks, and scale what wins.
-
GEO Testing in Ecommerce: Moving from Guesswork to Proof: how GEO testing mirrors SEO testing, where PDPs and PLPs act as the testing ground, and what ecommerce leaders can do right now to prepare for AI-driven discovery..
Key takeaways
-
SEO now spans a whole ecosystem, not only your site. Mike treats search more like reputation management, where brands need visibility across many surfaces and formats, not only one ranking URL.
-
Query fan-out means more chances to win or lose. Large language models break a single query into many related sub-queries. Each one is a raffle ticket, and the more of them you rank for with the right content type, the more often you appear in synthesized answers.
-
GEO is closely related to SEO but deserves its own frame. Tactics often overlap, yet the channels behave differently, serve different roles for brands, and require their own language, metrics, and expectations at leadership level.
-
SEO and GEO tests will not always agree. The same on-page change can help traditional blue-link SEO and hurt LLM visibility, or the reverse. That future demands multi-metric testing rather than single-metric decisions.
-
Agents and feeds will push work toward the data layer. As agentic experiences grow, machine-facing content, structured feeds, and quality controls will matter more, which is why new GEO tooling is already focusing on semantics and data-first architectures.
How SEO Has Shifted: From Pages to a Connected Ecosystem
I opened by asking Mike what has changed most in his work over the last few years.
His answer set the tone for the whole session.
SEO can no longer treat the website as the only surface. Teams now operate inside a content ecosystem: video, Reddit, social channels, owned content hubs, structured feeds, and anything else that might be retrieved or synthesized by a model.
Mike likens the shift to reputation management. In the past, reputation management meant securing multiple positions for a brand on a single low-volume query. Today, AI systems generate dozens of sub-queries, so the job is spreading influence across many keywords, many formats, and many surfaces.
You are not only competing for a blue link. You are competing for presence in the fan-out patterns that shape the models’ final answers.
What Query Fan-Out Really Is
Query fan-out is still widely misunderstood, so we dug into how it actually works.
Mike explained that systems like Google and ChatGPT do far more than run the original query. They:
- Look at related queries other users have searched
- Map the query into vector space and expand the intent
- Run speculative sub-queries to fill in gaps
- Select across a pool of 5 to 40 candidates (or sometimes more)
- Fetch the matching content type (lists, videos, guides, specs, comparison data)
The model then uses this evidence inside the synthesis pipeline.
Mike used a simple analogy: a raffle.
Every piece of content you have that ranks for one of these fan-out queries is another raffle ticket. Once the model enters the synthesis stage, you no longer have control. Your influence happens earlier, when you earn those tickets.
This is why GEO work must target not only broad keywords but also the long tail of evidence the model retrieves.
Why Google May Have a Structural Advantage
We compared notes on Google’s position in AI search. I am more bullish on Google than some commentators, and Mike agrees.
The reasons:
- They own their own chips and can scale model compute with fewer constraints.
- They have multibillion-user platforms feeding training and usage signals.
- They can run large batches of fan-out queries cheaply inside their existing index.
- They have decades of spam-fighting experience and quality systems.
- They can integrate personal context from Gmail, Maps, YouTube, Chrome, and others.
ChatGPT has innovated quickly, but it still lacks Google’s infrastructure, index depth, and adversarial experience. These factors matter when models operate at global search scale.
Why GEO Needs Its Own Name
We turned to the naming debate: GEO, AIO, LLM SEO, or “just SEO”.
Mike’s view is pragmatic. Senior leaders in enterprise companies are calling it GEO, and the name came directly from the research paper that first described the field. More importantly, telling executives that this is “just SEO” closes doors.
Even if some tactics overlap, the channel is different.
These systems weight evidence differently, surface content in new ways, cite sources unpredictably, and blend inputs from feeds, APIs, SERP features, and raw pages.
Mike gave a useful analogy: the tactics of cold outreach, PR, and link building share similarities, yet nobody treats them as the same job. GEO deserves its own frame, language, and responsibility.
Why GEO Tests and SEO Tests Won’t Always Align
We discussed whether SEO tests and GEO tests are identical.The answer is no.
Even when the on-page change is identical, the channels may react differently.
Mike gave an example:
Removing meta descriptions can improve CTR in classic SEO, but ChatGPT may rely on meta descriptions that include a “spoiler” of the answer. In that case, removing them hurts your presence in LLM responses.
We both expect to see more cases like this:
- Positive for blue-link SEO
- Negative for LLM performance
- Or the reverse
That future requires multi-metric testing, not single-metric decisions.
How Agents May Change Search: From Many Customers to One
We also explored the idea of agents as the next layer of search.
James, the founder of Profound, has described agents as a world where brands market to “one customer” - the agent.
My take is that agents will be as individual as the users they represent.
Mike’s take is that these agents will still retrieve from the same internet, but the filtering will become far more specific. That specificity could produce:
- Hyper-narrow content variants
- Large volumes of near-duplicate material
- A stronger need for clear data structures
- A shift from front-end presentation to data layer clarity
The role of feeds, protocols, and structured attributes will grow. Canonicalization may evolve into a dual system: one for humans, one for machines.
Why Model-Facing Content Changes the SEO Playbook
We also touched on an important implication of agentic commerce.Some systems actively encourage publishers to provide machine-formatted versions of their content. This opens the door to misuse, such as over-stuffed content only meant for bots.
Without guardrails, the risk of spam is high. Google’s long history of adversarial thinking gives it an advantage here. Other platforms may need to evolve quickly.
The Future of GEO Tools
Mike has worked closely with the Profound team and others building GEO tooling.
He highlighted two themes:
- Speed of innovation - GEO tools ship features faster than traditional SEO platforms.
- A reset of outdated assumptions - GEO tools start with semantic understanding, not lexical scoring or old heuristics.
This new tooling space may end up defining technical standards much like crawlers, log analysis, and rank trackers shaped earlier eras.
How Artists Are Adapting to AI (and What SEO Can Learn)
We ended on something personal. Mike is an artist as well as an engineer. He uses LLMs as feedback tools and sees them as extensions of creativity, not replacements. Every creative field goes through this cycle: initial pushback, then adoption once a respected creator embraces the new medium.
The lesson applies to our industry too. AI will not erase SEO, but it will reshape how we practice it.
Put Search in Control Mode with SearchPilot
Search is often the biggest channel and the least understood. SearchPilot makes SEO (and now GEO!) testable so leaders can move from guessing to knowing.
We run controlled experiments across category pages, product detail pages, navigation, and content, then deliver clear uplift with timelines and confidence. Teams progress from quick validation to a steady test cadence to full control, turning search into a performance channel you can plan and fund.
For ecommerce teams focused on product grids, Merchant Center feeds, and variant handling, the first step is a focused test plan. Measurement tracks impressions, clicks, and revenue so leaders can see the real impact.
Stop trying to predict the future. Experiment to discover it. If you want tailored test ideas for your top PLPs and PDPs, schedule a demo and we’ll share a starter list and a clear path from validation to velocity to control.
