How AI Overviews, AI Mode, ChatGPT search, Perplexity and Copilot select sources — and what you can change.
Delivered by the AISEOExpert team, who do the work themselves.
Method stages this service covers
AI search optimization is the work of making a company findable and accurately described in AI-generated answers: Google's AI Overviews and AI Mode, ChatGPT search, Perplexity, Copilot and Gemini. It rests on the same foundations as SEO, because these systems retrieve from the web index, and adds three things: a check that AI crawlers can reach your pages, an entity layer so systems know who you are, and content written so a passage can be lifted and cited. We measure it by sampling prompts and recording mentions, citations and accuracy, with the run-to-run spread shown.
The problem
Buyers ask assistants to summarise a category, compare vendors and recommend a shortlist. Your company is missing, described wrongly, or represented by a competitor's page — and no one on the team can see it happening.
We baseline how assistants currently describe you, fix the access, entity and evidence causes, and measure the change with a published prompt-sampling method.
Assistants recommend competitors for the category you lead.
Old positioning, wrong pricing or a product you no longer sell shows up in answers.
Comparison pages written by others are the source assistants quote about you.
AI referrals show in analytics but nobody knows which prompts or pages drive them.
Assistants answer from a mix of model knowledge and live retrieval. Retrieval depends on the same things search does: pages the crawler can reach, content that is clear about what the company is, evidence that makes a page worth citing, and passages that can be lifted as an answer. That is why this service runs across five stages of the Crawl-to-Citation Method. SEO vs AEO vs GEO explains how the labels relate and which one a given symptom needs.
We run the same measurement on this site and publish the results in the Lab, including the flat months.
How we approach it
We measure first, then fix what blocks retrieval, then what confuses description. Each step ends with a check we can show you.
See the full methodA fixed prompt set, run on a schedule across named engines, recording mentions, citations, accuracy and the competitors named.
Access, understanding, evidence and retrievability are checked in order. Most misses trace to one of them.
Entity corrections, structured data that matches the page, sourced answer-first passages, and pages for unanswered prompts.
Re-run the prompt set, report what changed and what didn't, and adjust.
What you get
Every finding names the engine, the prompt and the page, so your team can see what to change.
Mention rate, citation rate and accuracy across named engines
How we measure it
We record a baseline on a fixed prompt set before any change, then re-run the same prompts on the same engines afterwards, with the spread shown.
Source: Sampled AI prompts across named engines
Source: Sampled AI prompts across named engines
Source: Sampled AI prompts, scored against the entity sheet
Source: Analytics and CRM
Proof
Each result states the baseline, the time window and what we can't attribute to our work.
Google clicks per month
28 before, 129 after.
Source: Search Console, web search · Jan vs Sep 2026
It covers all six method stages, so AI-visibility gaps are ranked against everything else before you commit to a sprint.
Book the diagnosticKeep exploring
Straight answers to what buyers ask us most.
Still have questions? Book a callThe engines and model versions are named in every report. The default set covers Google AI Overviews and AI Mode, ChatGPT search, Perplexity and Copilot.
By sampling: a fixed prompt set, run several times per engine each month, with mentions, citations and accuracy recorded. The method is published in the Lab.
Google states that AI Overviews and AI Mode need no special files or markup. We treat llms.txt as an optional experiment, not a fix.
AI search optimization is the umbrella service. Generative engine optimization focuses on citable content; answer engine optimization on question-led pages. Many engagements combine them.
Last reviewed