Our sales team keeps hearing that prospects shortlisted vendors by asking ChatGPT, and we're never on those lists. What could be going on?
Sales keeps hearing that prospects shortlist vendors in ChatGPT and you are absent. The causes we see most often and how we diagnose them for clients.
This conversation starts the same way almost every time. Someone in sales says a prospect mentioned they asked ChatGPT for vendors, and the company was not named. The marketing lead wants to know if it is true, how bad it is and what to do. Here is how we work through it with clients.
Treat the anecdote as a brief
The best prompts we ever get come from sales calls. We ask for the exact wording prospects used, the company type they were, the country and what they were trying to solve. Those become the first prompts on the panel. We then add category, comparison and alternatives questions around them.
Then we test properly. One chat is not a test. In Promptwatch we run the panel daily per model, country and language and look at Visibility Score, Share of Voice, Sentiment and Position in Answer, next to three to five named competitors. If the client is truly missing, it shows within days. If they are present in some segments and not others, we see that too. Persona stacking helps here, because a CFO and an engineer ask differently.
The causes we see most often
Crawler access
Before anything else we check whether OpenAI's bots can fetch the site. GPTBot, OAI-SearchBot and ChatGPT-User are separate, and a bot protection rule or CDN setting can block them. We have found accidental blocks that nobody on the client side knew about. Agent Analytics shows IP-verified crawler logs, errors with explanations and a citation rate per page. Crawler logs start on the Professional plan, and we tell clients that up front so they can plan.
Missing answers to the sub-questions
ChatGPT splits a vendor question into narrower searches: pricing, integrations, security, a particular industry, comparisons. If the client's site has no clear page for those, there is nothing to cite. Content gap analysis with coverage scores shows which sub-questions are unanswered, and we start with the ones closest to revenue.
Third-party absence
This one surprises people. ChatGPT often draws on comparison articles, review sites, forum threads and videos. If competitors are named in those and the client is not, the client's own content cannot fully fix it. Citation analytics lists the pages and domains cited for each prompt, including Reddit, YouTube and LinkedIn reports, plus offsite mentions. We use it to build a short list of third-party places worth earning a mention.
Weak or inconsistent positioning
If a site says one thing, a profile says another and press says a third, the model has a muddier picture. We check how the brand is described in answers, using the Brand Book to set what counts as a mention, and we align the description where we can.
Segment gaps
A client may show up for small companies and not for enterprises, or in one country and not another. Segmenting by topic, persona and country shows it. State and city tracking begins on the Business plan, which matters for local vendors.
What we do next
The Action Board turns the evidence into a prioritized list, and we review it with the client weekly. The first month usually looks like this: fix access, close two or three high-value content gaps, plan third-party presence and set up visitor analytics so AI-referred visits and conversions are counted from the start.
We tell sales what we found and ask them to keep sending us prompts. They are a free source of real buyer language.
What we are honest about
We do not promise that a client will be on the list next month. ChatGPT answers vary, and competitors respond. Some fixes show up quickly and some never do. What we can promise is a clear picture of why the client is absent and a ranked set of things to try. We also tell clients that not every lost shortlist is caused by AI. Prospects ask colleagues, read reviews and search normally too. AI visibility is one input, and we want to measure it without overstating it.
If your sales team keeps hearing the same story, take it seriously, test it properly and start with access and sub-question coverage. Those two checks explain more than any clever tactic.