Get read, summarized, and cited by the AI answer engines.
AI search optimization, the work some people call GEO, is the strategy for being the source ChatGPT, Perplexity, and Google AI Overviews reach for when someone asks about your category. We ground it in the actual research on what moves citations, and we are candid about the part most sellers skip: no one can guarantee a citation, and the traffic is smaller than the hype suggests.
> Structure for what engines reward,
> measure it, promise nothing.
A growing share of buyers now ask an AI tool first.
Before someone contacts you, they increasingly ask an answer engine to do the shortlisting. They type a question into ChatGPT, or Perplexity, or Google's AI Mode, which crossed a billion monthly users in its first year, and they read the synthesized answer instead of scrolling a page of blue links. Whether your business appears in that answer, and whether it is described correctly, is decided by what these tools find when they read your site. AI search optimization is the work of shaping that.
It helps to be precise about the term. The research community calls this GEO, generative engine optimization, after the first controlled study of it, led by a Princeton team and presented at KDD 2024. We usually just say AI search optimization because it is plainer, but they mean the same thing: making your content the kind of source an AI answer engine will read, trust, summarize, and cite. This page is the strategy. The hands-on page craft that carries it out lives on our Answer Engine Optimization page.
Here is the part we will not soften. This is a young, unsteady field. Citation sets are volatile, independent trackers have watched them swing on the order of 40 to 60 percent month to month, and the referral traffic that reaches most sites is still small. So we treat this as one measured layer on top of solid search foundations, not a magic tap. The upside is real, being cited and described accurately when a buyer asks about your category, but the honest framing matters more here than almost anywhere else we work.
What this work is, and is not
- It is the strategy for being cited by AI answer engines, grounded in real research.
- It is measured against your own category, page by page, over time.
- It is not a guarantee of citations or rankings. Nobody can promise those.
- It is not a traffic firehose. AI referral traffic is still a small, volatile slice.
- It is not a replacement for SEO. It sits on top of it.
We work from evidence, not folklore.
Most AI search advice is guesswork dressed as certainty. There is, though, one solid anchor: the GEO study from Aggarwal and colleagues, presented at KDD 2024, which ran a controlled experiment across thousands of queries to see which content changes actually raised how often a page got cited in an AI answer.
The findings are refreshingly concrete. The moves that worked were about adding real substance: citing genuine external sources, including credible direct quotations, and adding accurate statistics with their origin named. The move that flopped was keyword stuffing, the old SEO reflex, which produced no benefit and in some engines nudged things slightly worse. And the payoff was not uniform, the same tactic helped more in some topics than others.
That is the whole philosophy of how we work here, in one paragraph. Add attributable substance, skip the tricks, and expect the effect to vary by domain.
// substance is the tactic; stuffing is a relic
What the GEO study found
| Content move | Moved citations? | Why it fits how engines work |
|---|---|---|
| Cite real sources | Yes | Gives the engine a verifiable claim to lean on |
| Add credible quotations | Yes | Quotable, attributable, easy to lift accurately |
| Add accurate statistics | Yes | A concrete, sourced fact an answer can carry |
| Keyword stuffing | No | Adds no substance; engines are not counting phrases |
The honest caveat. This is one study, run on systems that keep changing, and its own authors found effectiveness varied by topic. We treat it as the best evidence we have, not gospel. It shapes where we start; the measurement decides what we keep.
The strategy engagement, spelled out.
AI search optimization is the umbrella. It decides which topics are worth targeting, how we measure whether you are being cited, and where the on-page effort goes. The actual page surgery, the definition blocks and comparison rows, is executed under Answer Engine Optimization; here we set the direction and keep score.
Everything below is sized to your category rather than run off a universal checklist, because the research is explicit that what works varies by domain. If a piece will not earn its keep for your business, we say so instead of billing for it.
What the engagement includes
- A curated prompt set for your category, the questions buyers actually ask
- A baseline read of how AI engines describe you now, and whether they cite you
- Citation share and citation accuracy tracked as distinct measures
- A priority map of the topics and pages worth the effort
- An evidence-based plan drawn from what the GEO research supports
- Coordination with the on-page rewrites done under AEO
- Re-tests on a schedule, so you watch the rate respond to real changes
- Plain-language reporting, including when something did not move
Measure first, change the substance, then re-measure.
The whole method is a loop. We find out how AI tools describe you today, add the kind of real, attributable substance the research rewards, then re-run the same prompts to see what actually moved. No guesses dressed as results.
// baseline, substance, re-test, repeat
- Baseline against your categoryWe build a prompt set from the questions your buyers really ask an AI tool, then record how ChatGPT, Perplexity, and Google's AI answers respond: whether you appear at all, whether you are cited, and whether the description is accurate. That is the starting line everything else is measured against.
- Pick the plays the evidence supportsWe choose where to focus using what the GEO research actually found moved citations, real sources, credible quotations, accurate sourced statistics, and skip the tricks that do not. Because effectiveness varies by domain, we weight the plan toward what fits your topic rather than a generic checklist.
- Add real, attributable substanceWorking with the on-page craft under Answer Engine Optimization, we make each priority page genuinely more citable: sourced facts, quotable statements, clean definitions, extractable structure. The visual design rarely changes; the substance and the structure do the work.
- Re-test and read it honestlyWe re-run the same prompts on a schedule and report what moved, what did not, and what is still noisy. Given how volatile citation sets are, we look at trend over several reads rather than reacting to a single month's swing.
- Decide what is worth continuingEvery so often we ask the blunt question: is this still earning its place for your business? For some clients the answer is yes, keep going; for others it is that the foundational SEO matters more right now. You get the straight version either way.
Frequently asked questions
Related services
AI search optimization is the strategy layer. The on-page craft that carries it out, and the search foundations it sits on, live alongside it. For the thinking behind how these two fit together, read GEO vs. SEO in our Insights.
Want to know how the AI answer engines describe you today?
Send us your URL and a sentence about what you sell. We will run a baseline across ChatGPT, Perplexity, and Google's AI answers, tell you whether you are cited and whether the description is accurate, and give you an honest read on whether this work is worth it for you yet.