"What should I check before signing with this vendor."
"How easy would it be to migrate off this platform later."
These are the kinds of questions buyers actually type into ChatGPT, and the answer they get back may have been written by a competitor, not by you.
AI needs criteria before it can answer
Ask AI whether a vendor is a good fit, and it isn't just retrieving a fact, it's making a judgment. A judgment needs criteria: what trade-offs actually matter in this category, what risks are worth raising, what separates a good fit from a bad one.
As an example, the evaluation criteria for enterprise loyalty software might include architecture fit, implementation complexity, governance, integration depth, and roadmap certainty. It's the same list a real positioning exercise produces, who the buyer is, what they're trying to avoid, what trade-offs they'll accept.
Most companies do this work, then keep it on their homepage. The criteria exist, but they rarely travel beyond the company’s own website.
Buyers ask AI what matters when choosing a vendor. If you have not explained how your category should be evaluated, AI will use the framework someone else has provided. In one study across B2B software categories, competitor-published content appeared in 70 percent of AI answers to buyer-style questions. [1]
Your Competitor Explains What Buyers Should Look For
A guide from Voucherify, an enterprise loyalty software vendor, shows what this looks like in practice. The guide compares fifteen platforms, including its own. Before discussing any vendor, the guide sets out the evaluation criteria: architecture fit, flexibility, governance and speed to market. Then it assesses each vendor.
Ask AI whether Fielo is worth considering outside Salesforce, and the guide has already framed the concern: it is “most compelling only inside Salesforce.”
Ask what may become difficult as Comarch scales, and the guide points to upgrade friction from customization, and support that “may become more contract-heavy over time.”
Ask how much control a buyer would retain with Kobie, and the guide frames the trade-off: its value “leans heavily toward a service-rich relationship,” rather than “a pure infrastructure play.”
Two vendors on the list had recently been acquired. The guide treats the acquisitions as questions buyers should investigate. Instead of treating the acquisitions as reasons to avoid them, the guide tells buyers what to ask the vendors, such as whether the product is still receiving active investment and how its roadmap might change after the deal.
The guide does not present any of these trade-offs as inherently bad. Every vendor also gets a green flag. Voucherify’s own entry notes that teams must build the customer-facing interface themselves.
Once an evaluation framework is published, it no longer helps only the company that created it. It also becomes one of the ways AI learns how to judge the category.
That raises another question: how much influence does that framework actually have on AI’s answer?
Citation does not tell you the impact
AI does not begin every answer with a fixed checklist for evaluating the category. It assembles an evaluation framework for the question in front of it.
That framework is shaped by three things: what the buyer asks, what the model already knows, and the current sources it draws on while forming the answer.
That is why citation alone does not tell you the full impact of a source. A conventional GEO analysis checks whether AI cites the guide. The deeper impact is that Voucherify’s evaluation framework is shaping how AI judges the category.
A competitor does not need to say, “Do not buy this vendor.” It can define what a careful buyer should examine. If those criteria begin to shape the answer, every vendor may be judged against them, including vendors the competitor has never mentioned directly.
AI Learns from Everyday Discussions Too
A competitor does not need to publish a detailed comparison guide to shape how AI answers buyer questions. Public conversations can have the same effect. In one account, a company published forty blog posts over a year.[3] None of them were cited when prospects asked AI how to choose a platform in that category.
A competitor with almost no blog spent roughly sixty-five minutes a week answering questions on Quora, joining Reddit discussions and replying to comments on YouTube reviews. When prospects later asked AI how to choose in that category, the competitor’s Quora answers were cited twelve times.
The competitor was not comparing itself with other vendors. It kept answering the question buyers were already asking: how should I evaluate this kind of platform?
One approach is a structured comparison guide. Another is answering buyer questions in public. Both help shape what AI considers important when evaluating the category.
The Framework AI Learns From
Every positioning exercise produces more than messaging. It produces a view of what buyers should evaluate, which trade-offs matter, and which risks deserve attention. The question is whether that thinking ever becomes public.
A comparison guide is one way to publish it. Answering buyer questions in public is another. The format seems to matter less than the outcome. Buyers and AI can only draw out of the evaluation frameworks that exist outside of your company.
The competitive question is no longer just whether your company appears in AI answers. It is whether your view of how the market should be evaluated becomes a part of the answer.
Do you know which evaluation criteria AI is using to judge your company?
Request an AI Search Visibility AuditReferences
- Hard Numbers, "Gullible Engine Optimisation," 2026.
- Voucherify, "Best enterprise loyalty software in 2026: 15 platforms compared," June 2026.
- Carlo Cuman, "AI Citations: 95% Come From Sources You Don't Own," Medium, March 2026.