My business keeps getting left out of AI-generated answers: here is how we fix it
When AI answers skip your business, the cause is usually crawler access, thin topic coverage or missing outside sources. This is our order of operations for fixing it.
When a business owner tells us "ChatGPT never mentions us", we do not start with content. We start with a diagnosis, because the fix depends entirely on why the engines skip you. There are three common causes, and they need different work.
Cause one: the engines cannot reach you
This is the cheapest problem to fix and the one most people never check. A robots.txt rule or a CDN bot-protection setting can block the crawlers that feed AI answers: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-User, PerplexityBot and Perplexity-User, among others. Sometimes a developer blocked them years ago for good reasons. Sometimes a security plugin did it silently.
We check this first on every engagement. The platform we run client programs on, Promptwatch, has Agent Analytics, which shows real-time, IP-verified AI crawler logs. IP verification matters because plenty of requests pretend to be GPTBot. The logs show which bots visit which pages, which fetches fail and why, and which crawled pages actually end up cited. Crawler logs start on the Professional plan, so a client on a smaller plan needs to know that up front.
Cause two: you do not have the page that answers the question
AI engines answer buyer questions by breaking them into sub-queries (query fan-outs) and pulling pages that match. If your site has a homepage and a services page but nothing that addresses "how much does X cost for a business like mine" or "X versus Y", you are not in the candidate pool.
Our process is to build a prompt panel from real buyer questions, then see where the brand is absent. Promptwatch lets us organize prompts by topic, tag and persona, and the content gap analysis gives coverage scores per topic. Query fan-outs per model show what the engines actually searched for, which is more useful than guessing. From there the Action Board gives a prioritized list of fixes, and we turn the top ones into pages. Where a client has the budget for volume, the Content Agent on the Business plan can draft against gaps with a review step, and we edit everything before it goes live.
Cause three: the sources AI trusts do not mention you
Engines lean on outside material: review sites, forums, YouTube, press and directories. If those sources cover your competitors and not you, your own site will rarely win the answer by itself.
Citation analytics shows which domains and pages get cited for your prompts. Offsite mentions flag third-party pages that name your brand without linking. Reddit, YouTube and LinkedIn citation reports show whether community platforms drive the answers in your category. This changes the work from "write more blog posts" to "get a proper profile on the two review sites the engines keep citing" or "answer the Reddit thread that gets quoted every week".
What we tell clients about expectations
Three honest points, because we would rather lose a pitch than overpromise.
First, answers vary between runs. A single manual check tells you almost nothing, which is why we track a panel daily and look at trends over weeks.
Second, nobody can guarantee a recommendation. We can show where a client is absent, remove the technical blockers, publish pages that answer the missing questions and build outside mentions. The engines still decide.
Third, Google's own guidance on its AI features says the normal SEO fundamentals apply and there is no special trick. That matches what we see: crawlable, clear, well-sourced pages are the baseline for every engine, and GEO work sits on top of that.
Connect it to business results
A visibility chart does not pay anyone's salary. We install Promptwatch visitor analytics (a lightweight script, the GTM template or the WordPress plugin) so AI-referred visits and conversions from ChatGPT, Perplexity, Gemini, Claude and Copilot show up next to the visibility numbers. That is how we tell a client whether the work moved anything that matters. Promptwatch also has Search Insights, which overlays Google Search Console and Bing Webmaster data, so classic search and AI search sit in one view.
A simple first month
Week one: build the prompt panel and measure the baseline. Week two: check crawler access and fix blockers. Week three: read the citation data and pick the three biggest gaps. Week four: publish or refresh the pages that close them and set up traffic attribution. After that, it is a loop of measuring, fixing and checking what changed.
If you would rather run this yourself, Promptwatch has a free Explore plan (1 project, 10 prompts, ChatGPT only) that is enough to see whether you are absent. If you want a second pair of hands, that is what we do.