Generative Engine Optimization Services

How the baseline works

The baseline is the foundation of the engagement, and it is easy to do badly. AI answers vary between sessions, between tools and sometimes between users, so a single screenshot proves very little. I build the baseline to be repeatable:

1. Question set. Typically a few dozen to a few hundred buyer questions, depending on how many services, industries and markets you cover. They come from real sources: sales conversations, CRM notes, Search Console queries and the language customers use.
2. Tools. The questions are run in the AI tools your buyers are likely to use, such as ChatGPT, Perplexity, Gemini, Claude and Google's AI features, without being signed in to accounts that carry your company's history where possible.
3. Repetition. Priority questions are run more than once to separate stable patterns from random variation.
4. Recording. For each answer: mentioned or not, cited with a link or not, description accuracy, competitors named and sources cited.

The result is a baseline you can compare against later, with the limitations stated clearly. It is a directional measure of how AI tools treat your business, not an exact market share figure, and I report it that way.

What the fix backlog usually contains

Baseline findings turn into a backlog of specific changes. The categories are consistent across most businesses, even though the details differ:

• Access. Robots rules, firewall settings or rendering issues that stop AI crawlers or search crawlers reading key pages. Allowing or blocking each crawler is a business decision, and I explain what each one is used for before recommending changes.
• Entity clarity. Inconsistent company names, service descriptions, locations or founders across your site, profiles and directories. Organization structured data and a clear about page help here.
• Answer-ready pages. Key service and topic pages restructured so the direct answer, definition or recommendation comes first, followed by supporting detail.
• Missing content. Questions where AI tools cite competitors or publishers because you have no page that answers them directly.
• Outside sources. Directory listings, review profiles, partner pages and industry publications that AI tools rely on for your category, corrected or strengthened where you have legitimate influence.

Each item has an owner, an effort estimate and acceptance criteria, following the same format I use in a technical SEO audit.

Diagnostic examples from AI answer baselines

These are illustrative patterns that baselines commonly reveal, described generally rather than from a specific client:

• Right category, wrong details. AI tools place the business in the correct category but list services it stopped offering, taken from an old directory profile.
• Competitors cited for your expertise. On a topic where you have deep experience, answers cite a competitor's guide because your knowledge only exists in sales decks and proposals.
• Name confusion. A company with a common name is merged with an unrelated business in AI descriptions, because nothing on the site disambiguates it clearly.
• Blocked by accident. A security setting or robots rule added for another reason prevents some crawlers from reading the site at all.
• Strong in one tool, absent in another. Different tools draw on different indexes and sources, so visibility can vary widely between them.

Each pattern points to a different fix, which is why the baseline comes before any content or technical work. Changing things without a baseline makes it impossible to tell what helped.

How this relates to SEO and other AI search pages

Generative engine optimization does not replace SEO. Most AI tools that cite sources rely, at least in part, on web search indexes and crawled pages, so a site that is hard to crawl or poorly understood in search usually struggles in AI answers too. The engagement therefore checks search foundations first and refers deeper technical problems to an SEO audit where needed.

Each AI surface also has its own behavior. Google's AI Overviews and AI Mode draw on Google's own index and ranking systems. ChatGPT search and Perplexity retrieve web results in their own ways. For those surfaces specifically, I have separate pages on AI Overview SEO and ChatGPT search optimization. This packaged service sits above them and covers all the tools your buyers use, measured with one consistent question set.

For the underlying concepts, including how answer-first content and entity clarity work, see my generative engine optimization consulting page and the related work on answer engine optimization.

KPIs and connection to pipeline

I do not promise that AI tools will mention or cite your business, because each tool decides its own sources and its behavior changes often. What I measure is whether the conditions and the outcomes are moving in the right direction:

• Mention rate across the question set, by tool and by topic
• Citation rate, where your pages are linked as a source
• Description accuracy, scored against an agreed fact sheet
• Competitor presence for the same questions
• Referral sessions from AI tools in GA4, where the referrer is passed
• Inquiries that mention AI tools as the discovery channel, captured in a CRM field or form question
• Progress through the fix backlog

The last two matter more than they look. AI tools do not always pass referral data, so many businesses cannot see AI-influenced leads at all. Adding a simple discovery question to forms and a matching field in the CRM, something I set up regularly in CRM consulting, gives a direct signal that analytics alone will miss.