The deal was decided in a place you can't see
Here's how buying actually works now. A procurement analyst or a manager on the buying committee opens ChatGPT or Perplexity and types something like "top vendors for X that support SSO and cost under Y per seat." They read the answer. They ask a follow-up. Twenty minutes later they have a working shortlist of four or five names, and they build the RFP around those.
By the time a formal RFP lands in your inbox, if it lands at all, the real filtering already happened. Quietly. In a chat window nobody logged. If your brand didn't come up in that conversation, you weren't rejected in a fair fight. You were never in the room.
And your sales team has no idea. There's no lost-deal notification for a deal you never heard about. No CRM record of the opportunity that got scoped without you. The pipeline just looks a little thinner than it should, and nobody can say why.
Being in the category isn't the same as being on the shortlist
A lot of brands assume they're fine here because when they ask an assistant "who are the top tools in our category," they show up. Good. But that's the easy prompt. That's not how a procurement person searches.
A buyer doesn't ask an open category question. They ask a loaded one. "Best vendors for a 200-person company in healthcare that need HIPAA compliance and Okta." "Which of these actually has a published API." "Cheapest option that still does role-based access." Every clause is a filter. Every filter is a place you can silently drop off the list.
This is the gap that kills you. You can be the third name in a generic category roundup and completely absent the moment someone adds the two or three constraints that define a real deal. The assistant isn't lying. It just doesn't have a clean, current reason to keep you in once the criteria tighten, so it reaches for the vendors whose specifics are easy to confirm.
The number that should worry you
22%
In this illustrative view of your dashboard, the brand appears in most broad category prompts but makes the shortlist on only 22% of the specific evaluation-criteria prompts a procurement team would actually type. The broad win hides the loss where the money is.
Check the prompts a buyer would really ask
You can't influence a conversation you can't see. But you can reconstruct it. The exact prompts a procurement team uses aren't a mystery. They're your category plus your evaluation criteria: compliance, integrations, price band, company size, industry, deployment model. That set is finite and knowable.
Crescive runs those prompts across ChatGPT, Perplexity, Gemini, Claude, Copilot, and Google's AI Overviews and shows you where you make the shortlist and where you vanish. Not just the friendly category question, but the constrained ones that map to how deals actually get scoped. When you drop off, it traces why: a missing spec on a page an assistant trusts, an integration you support but never documented in a way a model can cite, a compliance certification that lives in a PDF nobody crawls.
Then it drafts the fix, you approve it, and it re-checks the same prompts to show whether you moved from absent to named. That's the whole point. Not a vanity ranking, but evidence that you're now showing up in the quiet shortlisting that happens weeks before an RFP, back when the deal is still winnable.
Key takeaways
- The vendor shortlist is increasingly built in an AI chat before any RFP is written, so a brand can lose a deal it never knew existed.
- Ranking well for broad category prompts means little if you disappear the moment a buyer adds real evaluation criteria like compliance, integrations, or company size.
- Audit the specific criteria-based prompts a procurement team would type, find where you drop off, fix the underlying gap, and re-check to confirm you're back on the list.
FAQ
How do AI assistants influence procurement before an RFP is sent?
Procurement analysts and buying committee members use assistants like ChatGPT and Perplexity to build an initial vendor shortlist by asking category questions loaded with specific criteria, such as required integrations, compliance certifications, price range, and company size. The RFP is then written around the handful of vendors that came up. A brand that isn't named in that early conversation is filtered out before any formal process begins, often with no record that the opportunity ever existed.
How can a brand see if it's being left off AI-generated vendor shortlists?
Reconstruct the prompts a procurement team would actually use, which are your category combined with real evaluation criteria, and run them across the major assistants to see where you're named and where you vanish. Crescive does this automatically, tests the constrained criteria-based prompts rather than just broad category questions, traces why you drop off, drafts fixes behind a human approval gate, and re-runs the same prompts to prove, under Crescive's automated criteria, whether you moved from absent to shortlisted.