Submitted by Adrien Vermeirsch, Founder @ Enlightn
The supply of verified humans is shifting away from surveys. Market research is not ready.
Sample suppliers are not really survey companies. They are audience businesses. They invest to collect verified humans (emails, app downloads, registrations) and monetize that audience through whatever tasks generate the most revenue per user. The operating principle is the same as for any audience business: maximize Customer Lifetime Value per user.
And today, for an audience owner sitting on a pool of verified humans, surveys are arguably one of the worst options on the table to maximize CLV.
Two adjacent markets are pulling verified humans away from surveys
↳ Paid offers, games, and online tasks. Revenue per user is lower than surveys, but there are no qualification criteria, no quotas, no quality reconciliation, no disqualifications. The user knows what they signed up for. Every click generates revenue.
↳ AI training tasks. RLHF, benchmarks, model evaluations. These pay significantly more than surveys, both for platforms and for users. The demand is concentrated in a few frontier labs today, but there is no real doubt it will keep growing. High-quality human data is core to AI model training, and a premium pool of users is being recruited and retained around it. Prolific just posted a $10M revenue week, at roughly the same time they doubled down on AI training tasks.
When demand evolves, supply adapts. Volumes in these adjacent markets are higher than ever, and they are sharing the same finite resource: verified humans. Surveys are being cannibalized by higher-yield ways for audience owners to monetize their users. That is why we see fewer and fewer pure panels, and more sample suppliers running multiple revenue streams in parallel.
The math against surveys
Paid offers and tasks generate revenue at every click. Smooth experience. Low friction. Manageable churn.
AI training tasks pay more, attract a different and often higher-quality audience, and offer a credible long-term monetization story for users: earn meaningful money doing meaningful work.
Surveys, on the other hand, disqualify users roughly 2 out of 3 times. That is two-thirds of clicks generating zero revenue. Worse, each disqualification is a friction point. Frustrated users churn. A user who could have generated CLV through other channels leaves because of the bad experience surveys gave them.
I think we are reaching the point where the average CLV of a sample supplier is being negatively impacted by survey opportunities. A senior leader at a sample supplier told me recently that participant churn from poor survey experience, combined with rising acquisition costs, was becoming unsustainable for them.
Which makes our industry’s reflex of asking suppliers to “do better on quality” increasingly disconnected from their actual economics. The most useful thing a high-quality supplier could do today is stop pushing their best participants into surveys. The most useful thing a low-quality one could do is keep fooling buyers with their marketing.
What happens next
When the supply of verified humans shifts toward higher-yield monetization, the supply remaining to absorb survey demand will skew toward the not-so-verified. The exact population already responsible for the data quality crisis we keep talking about.
I think we all agree we don’t want to reach that point.
Some will argue that synthetic data will be mature enough by then to replace real humans for surveys. I personally don’t see synthetic data as a replacement. I see it as another tool researchers use for the right kind of study. It will absorb low-stakes work. It will not absorb the studies where you genuinely cannot afford to be wrong.
So what do we do from there?
I think the focus has to be on increasing the survey qualification rate. That is the single variable dragging down sample-supplier CLV from surveys. Increase it, and retention goes up. Revenue per click goes up. The economics of surveys as a monetization channel for high-quality audiences start to make sense again.
And to be clear, pre-screening is not the answer. By the time a user reaches a pre-screener, they are already in the survey-taking flow. The disqualification damage is already done. They clicked, they did not get paid, they walked away frustrated.
Solving it before the click is the only way we keep owners of high-quality audiences willing to monetize them through surveys, instead of watching them quietly move that audience to higher-yield options.
