What is generative engine optimization?

GEO is the practice of getting a business named, described accurately, and cited inside the answers ChatGPT, Perplexity, Gemini, and Google AI Overviews write. This page explains the trade in plain terms: the five names it goes by, the prompts it targets, how it is measured, and what it costs.

By Austin Brewer, founder of AI Visibility Partners · Updated 24 August 2026 · 11 min read

What is generative engine optimization?

Generative engine optimization (GEO) is the practice of getting a business named, described accurately, and cited inside the answers AI assistants write. The same work goes by four other names: answer engine optimization (AEO), LLM SEO, AI search optimization, and, when the target is an AI agent that acts rather than merely answers, agentic search optimization (ASO). They describe one trade. The difference from traditional search is structural. SEO competes for a position in a list of ten links. GEO competes for a sentence inside a single synthesized answer, where there is usually room for three or four names. The unit of success is a citation, not a ranking.

If a vendor uses one of these terms and you use another, you are still talking about the same thing. We use GEO throughout this site because it is the term most buyers now search for, but nothing on this page changes if you prefer AEO.

Why does AI visibility matter for a remodeling or design firm?

A remodel is a high-consideration purchase, and a growing share of that research now begins inside an AI assistant rather than a results page. In the largest published study of the problem, 78% of independent local operators had effectively zero AI citation share, while directories and national chains took the answer space. On the search side, organic click-through rates fell 61% on queries where an AI Overview appears, from 1.76% to 0.61%. The unusual part is that in most local markets the recommendation slot is not yet contested by anyone.

One honest caveat about that study, which most pages quoting it leave out: its six categories were plumbing, electrical, personal injury law, family medicine, dentistry, and HVAC. Remodeling and interior design were not among them. The structural finding — that directories and chains absorb the citations independents do not claim — travels well to adjacent high-consideration local services, but the 78% is not a remodeling number. Treat it as the shape of the problem rather than your firm's score. Your actual score is what an audit measures.

Engagement data points the same way. Adobe Digital Insights found AI-referred visitors converted 31% better than non-AI traffic, spent 45% longer on site, and bounced 33% less often, across US retail over the 2025 holiday season. That is retail rather than home services, so treat the direction as informative and the magnitude as unproven in this vertical.

Which AI engines answer these questions, and which ones get tracked?

Homeowners ask these questions across a widening set of assistants: ChatGPT, Google AI Overviews and Google AI Mode, Perplexity, Gemini, Claude, Microsoft Copilot, Bing Chat, and Grok. AI Visibility Partners tracks four of them as the standing measured set: ChatGPT, Perplexity, Gemini, and Google AI Overviews. Those four are tracked because they carry the bulk of US consumer research traffic in this category and because each can be measured repeatably from a clean session. The others — Claude, Microsoft Copilot, Grok, and Google AI Mode among them — are checked opportunistically and reported when something moves, but they are not part of the scored citation share.

This matters when comparing vendors. A tracked set of four engines measured the same way every cycle produces a number you can trend. A dashboard claiming coverage of a dozen assistants usually means sampling, not measurement. Ask any vendor which engines are scored, how often, and whether the method is identical between cycles.

What does an AI engine actually read before it recommends a firm?

An engine builds its answer from pages it can fetch, parse, and quote in a sentence or two. That favours a particular shape of content over a particular quality of prose. Five things decide it: whether the page can be retrieved at all, whether each section opens with a direct answer, whether the firm states plainly what it does and where, whether structured data spells out the facts, and whether anyone other than the firm's own website corroborates it. Pages written to persuade a human scrolling slowly tend to fail all five at once.

The five signals, in the order they typically cause failure

1. Retrievability. If AI crawlers are blocked in robots.txt, or the content only appears after JavaScript runs, nothing else matters. This is the most common single point of failure and the cheapest to fix.
2. Direct-answer structure. A question-shaped heading followed immediately by a standalone 50 to 150 word answer gives an engine a passage it can lift without editing. A section that opens with background does not.
3. Explicit scope. "We serve homeowners across the greater Austin area on kitchen and whole-home remodels" is usable. "Crafting spaces that inspire" is not.
4. Structured data. JSON-LD schema — LocalBusiness or ProfessionalService, Service, and FAQPage — states the service, the area served, the price range, and the principal by name, so an engine reads facts instead of inferring them.
5. Third-party corroboration. Engines weight what other credible sites say about you. A firm mentioned nowhere but its own website is a firm with one source. The 5W study measured an eight-fold entity-strength multiplier on citation share, the largest single signal effect it recorded.

Freshness compounds with all five rather than substituting for any of them. Seer Interactive found that 75% of the pages AI engines cite had been updated within the previous year, and that updating an existing page beats publishing a new one: 72% of cited pages showed a recent update, while only 42% had been newly published in that window.

Which prompts do buyers actually type?

Buyers type the questions they would ask a knowledgeable friend, not the phrases firms wish they ranked for. In US residential remodeling and interior design the questions cluster into four stages of one decision, and an audit tracks roughly 25 of them, drawn from a standing 50-prompt library and then held fixed so every cycle measures the same thing. Below are real prompts from the standing library, unedited. If a vendor cannot show you the exact prompt strings it intends to track before you pay, it is selling a dashboard rather than a measurement. The list below is the same one used to build a client set, with the city swapped for yours.

Comparison and cost prompts are the most valuable real estate in the set, because those are the answers a homeowner reads before they have a shortlist. They are also the prompts most remodeling firms have never written a page for, which is why directories keep winning them.

How do you measure whether an AI engine recommends a business?

You fix a set of buyer questions, run them the same way every cycle, and record what comes back. The measurement unit is citation share: the percentage of tracked prompts where an engine names the firm. Because AI answers vary between sessions, one screenshot proves nothing. Each prompt runs three times from a clean, signed-out session with no memory or prior context, and the answer that repeats is the one logged, with a dated screenshot attached. A firm counts as cited on a prompt when it is named in at least two of the three runs.

The four rules that make the number repeatable

1. Clean sessions only. Every prompt runs signed out, with no memory or account history to bias the answer toward a firm you have already looked at.
2. Modal answers logged. Three runs per prompt; the answer that repeats is recorded, not the most flattering one.
3. Screenshots attached. Every logged answer has a dated screenshot in the appendix, so the record can be checked rather than trusted.
4. Fixed prompt set. The set does not change between cycles. Movement in the number means movement in the market, not a new question.

The same method runs at baseline and at every re-measurement, which is the only way a before-and-after figure means anything. The audit page shows the full 25-prompt build and what the scorecard looks like.

What actually changes on a website during GEO work?

Most of the work is structural and invisible to visitors. Headings get rewritten as the questions buyers ask. The answer moves to the top of each section instead of following a paragraph of background. JSON-LD schema is added or corrected. Crawler access is fixed in robots.txt, and an llms.txt file is published. Cost and comparison pages get written, because those are the questions buyers ask that most firms never answer. A named author and an updated date go on the content. Visual design usually survives untouched.

Platform is rarely a blocker. WordPress, Squarespace, Wix, and Webflow can all carry the structure and the schema this work needs, though each has its own quirks about where code blocks are allowed. Google Business Profile and the local pack sit alongside this work rather than inside it: they still matter for maps results, and they are not what decides whether ChatGPT names you.

What does generative engine optimization cost?

At AI Visibility Partners a one-time AI visibility audit runs $500 to $900, and the fee credits toward the first retainer month. Ongoing work runs $750 to $1,200 a month on the Foundation tier and $2,000 to $3,500 a month on Growth, with the difference being content volume and citation outreach rather than measurement. Across the wider market, GEO retainers for small local firms generally sit in the same low four figures. Any vendor that will not state a range before a call is usually pricing off your reaction to it.

Full detail, including what drives a quote toward the top or bottom of each range, is on the pricing page.

Which firms is this worth paying for, and who should skip it?

GEO earns its fee when a firm already converts the leads it gets and simply is not being found: an established remodeling, design-build, or interior design business with real projects, real reviews, and a website written for humans rather than for retrieval. The spend is premature if the firm has no website worth citing, no capacity to take on more work, or a marketing budget small enough that one month of retainer would come out of materials. In that case fix the website basics first. GEO amplifies a foundation, it does not substitute for one.

We say no to firms in the second group on the call rather than after the invoice. The client roster stays deliberately small and inside one vertical, because benchmark data across similar firms in the same markets is worth more than a broad client list. The standing benchmark metros are Austin, Charlotte, and Denver.

Two disclosures worth reading before you call

There is no published case study yet. AI Visibility Partners was founded in 2026 and the first client engagement is in progress. Its results publish when the matched re-measurement window closes and not before, because a before-and-after figure taken outside a matched window has not actually been checked. If you need to see proven results in this vertical before you spend, that is a reasonable position and this is the wrong month to hire us.

The founder also runs a design studio. Austin Brewer founded Astratto Design, an interior design studio, as well as this practice. That is disclosed here rather than buried because AI Visibility Partners sells to interior design firms. If you want to know whether that creates a conflict in your market, ask about it on the first call, before anything is signed.

Where do these numbers come from?

Every figure on this page comes from a named third-party study, listed below. None of the numbers are ours. Full samples, the scope limits each study states, and the four commonly repeated statistics removed from this site rather than published with a vague attribution are all set out on the sources page.

1. 5W Public Relations, The Local Services AI Visibility Crisis 2026: national chains, platform aggregators, and the structural exclusion of independent operators, April 2026. View the study ↗
2. Seer Interactive, AIO Impact on Google CTR: September 2025 Update, 4 November 2025. View the study ↗
3. Seer Interactive, Study: Content Recency’s Impact on AI Visibility in 2026, 24 July 2026. View the study ↗
4. Adobe Digital Insights, AI traffic analysis, January 2026. View the study ↗

Full sources, samples, scope limits, and the four removed figures →

What else do firms ask about GEO?

Where should you read next?

Three next steps, depending on what you are trying to settle: how this differs from the SEO you already pay for, what the work costs in detail, or how to tell a real vendor from a bad one.