How to Vet a Vendor’s AI or Data Claim Before It Becomes Part of Your Strategy

—The Active Listening Myth
Insight: How to Vet a Vendor’s AI or Data Claim Before It Becomes Part of Your Strategy
Author: IMB Editorial Team
IMB Journal – International Marketing Board
Volume 1 | Issue 9
September 2026

How to Vet a Vendor’s AI or Data Claim Before It Becomes Part of Your Strategy

The small businesses that bought Active Listening weren’t unusually careless. They were buying into a claim that felt plausible precisely because it matched a suspicion many of them, and their own customers, already carried about how modern advertising works. That’s exactly what makes this kind of pitch dangerous, and exactly why it’s worth having a specific, repeatable way to test a vendor’s claim before it becomes a line item in a marketing budget.

Three questions do most of the useful work here. First, ask the vendor to explain the mechanism in plain terms, not the marketing language, the actual technical mechanism, and notice whether the explanation gets vaguer rather than clearer as you press for detail. A real capability can usually be explained more precisely the more specific the question gets. A fabricated one tends to retreat into reassurance and testimonials instead.

Second, ask directly where the underlying data comes from and request documentation, not a verbal assurance. In the Active Listening case, the data was simply purchased from third-party brokers and resold at a markup, a fact that would have been discoverable by anyone who insisted on seeing the actual data source rather than accepting the vendor’s description of it. Any vendor unwilling to name its data sources, or unable to produce a contract or data-sharing agreement backing up the claim, is asking to be trusted on faith alone.

Third, and this is the question regulators themselves leaned on most heavily in this case, ask specifically how consumer consent was obtained, and whether that consent covers the actual use being claimed. A generic terms-of-service agreement for an unrelated app or device is not consent to a different company using that data for a different purpose, and any vendor who points to broad, indirect consent as the basis for a specific, sensitive capability is making the same move regulators rejected here.

None of these questions require technical expertise to ask. They require only a willingness to keep pressing past the first confident answer, especially when that answer happens to confirm something you were already inclined to believe. The vendors who can withstand this kind of questioning tend to welcome it, because they have real answers. The ones who can’t are telling you something important, well before a regulator ever has to.


This concludes our three-part series on the Active Listening case. Read Part 1, “When a Marketing Strategy Is Built on a Myth Customers Already Believe,” and Part 2, the Cox Media Group case study, in this issue.