The dirty secret of consumer research is finally being shouted from the timeline

For decades the market-research industry sold the same promise with the same smirk: ask a panel, weight the answers, ship the deck. The deck was a lie, and everybody working in the trade knew it. On 9 June 2026 the smirk finally got caught on camera, when Dom Wong, the founder of consumer-research firm Pogo, announced that the company had launched an AI product built to talk directly to verified buyers of the brands it studies — and to do so, in his words, in a market "overrun by fraud."

The point of the launch is not the chatbot. The point is the wreckage it has to clear. A survey ecosystem that lived off opaque panels, third-party verification stamps and self-reported demographics has been exposed, in industry data and trade press alike, as something between a rounding error and a confidence trick. Pogo is betting that brands will pay a premium for the one thing the legacy stack could not deliver: a respondent pool whose purchases can be confirmed, not inferred.

The product, such as it is

Pogo's pitch, as Wong laid it out on X on 9 June 2026, is straightforward. The product holds conversations with consumers whose status as actual buyers of a given brand has been verified — typically through transaction data, receipts or account linkage — and uses those conversations as the substrate for brand research. The AI is the front end; the verified-buyer graph is the moat. The video accompanying the announcement frames the rest of the industry as a sort of organised crime: panel farms, duplicate respondents, AI-generated survey answers, professional "survey takers" working four platforms at once.

None of those problems are new. What is new is that the platforms themselves have, over the last two years, been quietly automating them at scale. Generative AI can fill a 20-minute survey faster, and with more convincing open-ends, than any human panel farm. The verification layers that the legacy research stack added in the 2010s — captchas, ID checks, attention metrics — were designed to stop humans pretending to be other humans. They were not designed for a system in which the humans are optional.

The defence the industry will offer, and why it won't hold

The big research conglomerates will, predictably, argue that this is just another entrant with a fresh methodology, and that the established playbooks — probability panels, double-opt-in recruitment, post-stratification weighting — already account for fraud. That defence is getting harder to make. The internal estimates that circulate in trade press suggest that, on the worst-performing panels, double-digit percentages of completed responses are not just inattentive but synthetic. The data the brands act on is, at the margin, the output of language models trained on the brands' own previous campaigns.

There is a subtler defence, too, and it is the one Wong's product really attacks. The research industry's value-add, the argument goes, is not raw response integrity but the weighting — the demographic and behavioural corrections applied after the fact. The whole discipline has, in effect, conceded that the unweighted sample is garbage and that the pricing model is really a fee for statistical reanimation. If Pogo can start from a verified-buyer population and skip the resurrection step, the weighting industry has nothing left to sell.

Why the marketing stack should be paying attention

This matters well beyond market research. The same fraudulent dynamics contaminate creator-marketing audits, brand-lift studies, attribution panels, and the audience data that publishers sell to advertisers. The verification layer that the open web never built is being retrofitted, market by market, by whoever can credibly link a person to a transaction. Pogo is one entrant; the retailers, card networks and consumer-finance apps sitting on the same underlying data are the much larger threat to the existing order.

The structural shift here is the same one playing out in advertising more broadly: first-party, transaction-adjacent signals are displacing the third-party, behavioural proxies that dominated the 2010s. The platforms that monetised the proxies — research firms, ad-verification vendors, attribution specialists — are discovering, in sequence, that the underlying ground has moved.

The serious bit

There is a consumer-protection story hiding inside this launch that deserves more attention than the marketing-trade press will give it. A research industry that quietly accepted double-digit fraud rates was not just defrauding its corporate clients; it was training a generation of brand managers to make decisions on data they had been told, in writing, to trust. The downstream cost of those decisions is borne by everyone who buys a product that exists because a survey, somewhere, said they would. The honest version of Pogo's pitch is not "better research." It is: the previous version was not research at all.

The desk is sceptical of the marketing industry's ability to police itself; the same incentives that produced the present mess are still in place, and the verification moat is only as deep as the receipts behind it.