Why we pay respondents before the data is sold

There is a quiet assumption in a lot of research platforms: that the money to pay respondents will be there when the work is done.

Usually it is. The failure mode is rare and specific — a researcher abandons a study, a company changes direction, a budget is pulled — and when it happens the people who already did the work are the ones who absorb it. They answered the questions. The payment simply never arrived, and there is nobody left to ask.

Escrow changes who carries that risk

On Surdatics a reward pool is funded before a survey is visible to anyone. The money is committed at the point the survey goes live, not at the point someone asks to be paid.

This has a consequence worth being explicit about: a survey that cannot pay is never shown. If the pool cannot cover another reward, the survey does not appear in anyone's list. An empty list costs far less trust than a survey that takes ten minutes of someone's attention and then pays nothing.

What we learned building the payout side

Two things surprised us.

A payout method that cannot settle is worse than one less method. Offering a rail that fails after someone has picked it and had their balance debited is a specific kind of broken. So a rail is only offered when the treasury behind it can actually pay — checked against the provider, not against a configuration flag someone set months ago.

Minimums need explaining, not just enforcing. Moving small amounts costs more in fees than the payout is worth. That is a real constraint, but a threshold with no explanation reads as an excuse not to pay. So the minimum is shown with the reason, and with how far away you are from it.

The point

A respondent giving you fifteen minutes is extending credit. Escrow is how you stop asking them to.