GET INTO THE ANSWER. Then get picked.
AI search optimization is one question: when a buyer asks ChatGPT, Gemini, Perplexity, Claude, Copilot or Google AI what to buy, does the answer say your name? We make it say yours, and we count how often. 100 real buyer prompts, every engine your market uses, many runs. One number: your Share of Decision.
More buying now starts as a question to a machine than as a search. The answer names three brands, maybe four, and nobody clicks. Two in three Google searches already end without a click. Only about one cited page in eight ranks in Google’s top ten for the same question, so a page-one ranking is not a seat at the table. Whoever the answer names gets the customer. We make that you, and we make it stick. SEO, AEO, GEO, AI shopping feeds, entity authority: one practice, because the machine does not care which acronym you used.
100 real buyer prompts, four engines, three runs each, from your market. Reported as a range against each competitor you name and the ones we find.
The domains each engine cites when it recommends a competitor, and which of them never mention you. This becomes the Digital PR target list.
Organization and Product schema, AI crawler access in robots.txt, pages that render without JavaScript, and a feed that carries GTIN, brand and availability.
Every fix with how to verify it and how to roll it back. Titles, canonicals, internal links, page speed: the SEO underneath, done properly.
Pages written to be quoted: the answer in the first 60 words, the facts in the HTML, the proof beside it.
The same brand facts everywhere the engines look: your site, profiles, directories, data sources and the press. Accuracy flags fixed first.
Google Merchant Center and ChatGPT Shopping readiness, with the feed rules per platform and the waitlists tracked.
The same prompt set every month, so you watch one number move. First-party data outranks tool estimates; where they conflict, we say so.
Before we touch anything, we measure your Share of Decision: 100 real buyer prompts across five intents, four engines, three runs each, from your market. 1,200 answers. Reported as a range against each competitor, never a rank, because AI tools return the same list less than once in 100 runs.
Then this practice does its part of the loop: get you into the answer and keep you there. The same prompt set is rerun every month, our own agents do the measuring, and a senior tells you what moved and why. First-party data outranks tool estimates; where they conflict, we say so.
Share of Decision measured against each competitor. Schema, crawler access, feed and rendering fixes live. The first answer-first pages published. Citation-source map delivered to Digital PR.
Mentions and citations on the sources the engines already trust. Prompt set rerun; the first range reported against month zero. Shopping feeds live where the platform allows.
Share of Decision rising across ChatGPT, Gemini, Perplexity and Google AI, with the sources that cited competitors now citing you. One number in your board deck.
These markers are how we run the practice, not an industry benchmark; none exists. Feed and technical fixes move within weeks. Earned mentions take months, and they are what moves the number.
In 2024, AI assistants sent this electronics retailer zero sales. Today, AI search is a seven-figure sales channel for it.
“We went from a single-market Meta presence to a genuinely global acquisition engine.”
SOURCED WHERE A SOURCE EXISTS
Mostly SEO underneath. What is new is earning mentions on other sites, keeping product feeds clean, and measuring your share across many answer runs instead of one ranking. Google says its AI features need no special optimisation beyond good pages.
SOURCE: Google Search Central; Search Engine Journal
No. Google still sends most of the traffic, and AI answers draw from the same web. But 68% of US Google searches ended without a click between January and April 2026, so the click is no longer the whole game.
SOURCE: Search Engine Land, 2026 zero-click study
Because other sites talk about them. Brand mentions on third-party sites correlate 0.664 with AI Overview visibility; backlinks correlate 0.218. The engines trust what everyone else says about you more than what you say about yourself.
SOURCE: Ahrefs, AI Overview brand correlation study
Not reliably. Only about 12% of the URLs AI engines cite rank in Google’s top ten for the same prompt. Ranking gets you considered; being cited gets you named.
SOURCE: Ahrefs, AI search overlap study
No one can. AI tools return the same brand list less than once in 100 runs. What we commit to is a measured rise in your Share of Decision across 100 prompts, four engines and many runs, reported as a range you can check.
SOURCE: Search Engine Land, AI recommendation lists rarely repeat
By measuring many times. 100 prompts, four engines, three runs each, from your market, reported as the share of answers you appear in. Tool data is directional, 10% of prompts are re-run by hand, and the method is in every report.
SOURCE: Search Engine Land; PPC Land on LLM tracking accuracy
No. Google said in June 2026 that llms.txt neither helps nor hurts, and a study of 300,000 domains found no effect on citations. Ordinary Organization, Product and Article schema, kept accurate, is what the engines read.
SOURCE: Google Search Central updates; Search Engine Journal, 300k-domain study
Clean product feeds and a healthy Merchant Center: GTIN, brand, price and availability on every item. ChatGPT Shopping is US-only for now; Shopify catalogs are integrated and other platforms join a waitlist.
SOURCE: OpenAI, merchants programme
Ads in ChatGPT sit apart from the answer and are labelled as ads. OpenAI says they do not influence what the answer recommends. Paid placement is a Paid Media job; being recommended is this one.
SOURCE: OpenAI, approach to advertising
Mostly not as a referral. In B2B, about 8.8% of AI-influenced visits arrive tagged as a referral; about 56% come later through branded search. We report Share of Decision alongside branded search and GA4, and say which number is the source of truth.
SOURCE: Relevant Audience, ChatGPT referrals in B2B
Feed, schema and crawler fixes move within weeks; earned mentions take months. By day 30 the baseline is measured and the technical fixes are live. By day 90 the first earned mentions land and the prompt set is rerun. By day 180 the number is moving on all four engines.
SOURCE: AB Media Co practice; no neutral benchmark exists
It depends on how many markets, engines and pages are involved, so we scope it after a $1,500 Share of Decision audit, credited against month one if you go ahead. For context, US estimates run $1,500 to $5,000 a month for small brands and $5,000 to $25,000 for mid-size.
SOURCE: AB Media Co terms; Nutshell GEO cost estimate (vendor figure)
We use cookies to measure how the site is used and how our ads perform. Necessary cookies always run. Cookie policy
