Your next RFP was shortlisted by an AI before you were invited

The shortlist has changed hands

For fifteen years, preparing an RFP followed a familiar routine: a few Google searches, two or three peer referrals, an industry directory. The shortlist was built from whatever the buyer found.

Today, part of that work happens in a single question to an assistant: which firms are best for migrating an Adobe Commerce site in Canada? The assistant replies with three to five names, often with a rationale. The buyer may never look further. There is no page two.

For a vendor, the consequence is simple: if you are not in that answer, you are not in the race, and you will not know it.

Ranking is not being named

A search engine ranks documents by relevance and authority. An answer engine does something different: it composes an answer and must decide which companies it can name without risk of being wrong.

That changes the rules. Being named takes more than a well-ranked page. The AI needs to answer four questions with confidence:

1. Who are you? A single entity, with a stable name, an address and identified leaders.
2. What do you do? A clear category, described the same way everywhere.
3. What proves it? Named projects, clients, results attributable to you.
4. Who else confirms it? Independent third-party sources that say what you say.

A company can rank well on Google and fail all four. Many firms that optimized page by page, without ever working on their identity, are in exactly that position.

The cost of a blurred identity: our own case

We ran this diagnosis on Nuevamed before offering it to clients. It was instructive.

Searching for our name surfaces three labels: Nuevamed, Nuevamed Communications and Nuevacommunications. Industry directories describe us as a marketing agency, a development firm or a communications agency, depending on the source. Some profiles are years old.

To a person, these gaps are trivial. To a model deciding whether it is looking at one company or three, and what that company actually does, they create doubt. And a model in doubt does not cite.

So we started with the least visible work: one name, one description, the same facts everywhere.

How to measure your presence

Protocol. Thirty questions a real buyer would ask before an RFP, in three groups: choosing a firm ("best Adobe Commerce agencies in Canada"), choosing a solution ("B2B platform for a distributor with an ERP") and validation ("is Nuevamed reliable for a Shopify Plus project?"). Each question is asked to three assistants (ChatGPT, Perplexity and Gemini), three times, on different days, to account for variability.

Scoring. For each answer: companies named, their position, the rationale given and the sources cited.

The five levers that drive citation

1. One consistent entity. The same name, short description and facts on your website, your Google Business Profile, LinkedIn and the main directories. Declare former names explicitly in your structured data so models connect them to you.

2. Attributable proof. A case study that names the client, the problem, the solution and a measured outcome is worth more than ten service pages. AI cites what it can verify.

3. Consistent third-party sources. Press coverage, conference talks, professional associations, review platforms. Their consistency with your own claims matters more than their number.

4. Content that answers real buyer questions. Not visibility pieces, but direct answers to the questions that precede a decision: cost, timelines, risk, selection criteria. Assistants draw on sources that take a position.

5. Regular measurement. What is not measured is not managed. A fixed set of questions, run monthly, is enough to track your progress and your competitors'.

What this means for marketing leadership

Traffic is no longer a sufficient metric. A growing share of the decision forms without a single visit to your site. The metrics that matter become your presence rate in answers for your categories, the accuracy of what assistants say about you, and the share of inbound leads that mention an AI recommendation.

It also changes ownership. Brand identity sat with marketing, SEO with the web team, directories with nobody. They are now one workstream, with one owner.

Conclusion

Companies will not lose RFPs to AI. They will lose them before the RFP exists, in a shortlist they never saw. The good news is that the criteria are simpler than traditional SEO: be identifiable, be proven, be confirmed by others. They are also the criteria human buyers have always applied.

Nuevamed helps organizations with their SEO and AI search visibility strategy, from identity diagnosis to monthly measurement. To learn more about our team, visit our About page.