The incentive is the part of a study most often set by precedent — whatever the last one paid, adjusted for whatever the budget allows. It deserves more attention than that, because it is not paying for answers. It is buying back the people who would otherwise ignore you, and those people are the reason the sample means anything.
What the money is actually for
Not the response. A response is cheap and you can always get more of them by lowering the bar.
The incentive exists to close the gap between who would answer for free and who you need in the sample. Free responses come disproportionately from people with strong feelings and time to express them. The incentive brings in the indifferent middle, which is where most of your customers live and where almost none of your unsolicited feedback comes from.
Read that way, the question changes. It stops being "what is the minimum that gets responses" and becomes "what does it take to make somebody who does not care about us spend eleven minutes on this".
An unincentivised survey does not save money. It buys a cheaper sample, and the discount comes out of the part you were trying to measure.
The two failure modes
They pull in opposite directions and both are common.
Too low
Response rates fall, and they do not fall evenly. What remains is the enthusiastic, the aggrieved and the professional respondent — somebody for whom a low rate is still worth taking because they answer at volume and have learned to move quickly.
That last group is the quiet problem, because their answers look completely normal.
Too high for the task
Once the incentive is large relative to the effort, qualifying becomes the objective. People discover what the screener wants and answer accordingly, not usually as fraud but as ordinary optimism about whether they fit.
This is a screening problem, and the answer is a better screener, not a smaller payment.
The instinct to solve the second by cutting the incentive is understandable and wrong. It fixes the symptom by reintroducing the first, and you end up with a sample that is both unrepresentative and cheap.
Setting the amount
Work backwards from an effective hourly rate for the respondent's time, then multiply by how long the task honestly takes — including the screener, not just the survey.
| Profile | Effective hourly | 10-minute task |
|---|---|---|
| General consumer | $15 – $30 | $2.50 – $5 |
| Low-incidence consumer | $30 – $60 | $5 – $10 |
| Business decision-maker | $60 – $150 | $10 – $25 |
| Specific professional role | $100 – $300 | $17 – $50 |
| Clinical or highly regulated | $200 – $600 | $33 – $100 |
Two adjustments matter more than the base rate.
Incidence. If one in twenty people qualifies, nineteen of them spent time on your screener for nothing, and the cost of reaching one qualified respondent includes all of that. Low incidence raises the price of a study far more than the incentive line suggests, and why incidence dominates recruitment cost works through the arithmetic.
Effort beyond answering. Anything requiring the respondent to fetch something, install something, record something or wait — a diary entry, a photograph of a receipt, a screen recording — is not survey time and should not be paid as though it were.
Forms of incentive, and what each does
The amount matters and the form changes who accepts.
Cash or bank transfer is the most universally valued and the least likely to skew who participates. It is also the form people most trust, which matters when you are asking a stranger.
Gift cards are the common default. Slightly less valued than the equivalent cash, easy to administer, and they quietly exclude anybody who does not shop where the card is good — which in international studies can be most of a market.
Charity donation appeals to a specific and unrepresentative slice of people. Reasonable as an option alongside a real incentive, poor as the only choice.
Product credit works for existing customers and is close to useless for anybody else, including the churned customers whose views you most want.
Timing, which people underrate
Pay on completion, within days rather than weeks.
Slow payment is the single most common complaint in research participation, and its cost is not the complaint. It is that people who waited three weeks do not answer the next study, and the panel degrades in a way that shows up as declining response rates months later with no obvious cause.
The same applies to clarity. Say what the incentive is, in what form, and when it will arrive, before somebody starts. A study that discloses this up front fills faster than one paying more and saying less, which is a genuinely cheap advantage to hold. How payouts and timing work from the participant's side covers what that looks like in practice.
Screen-outs
The hardest part of the design and the one most often handled badly.
Somebody spends four minutes on a screener, does not qualify, and receives nothing. From their side that is indistinguishable from being wasted, and doing it repeatedly is how a panel learns to stop opening your invitations.
Keep screeners short
Two or three questions where possible. Every additional question multiplies the number of people you disappoint.
Screen before inviting where you can
Using what you already know about somebody is cheaper and kinder than asking them to prove it.
Pay a token for longer screeners
A small amount for a screen-out that took real time costs little against the study and keeps the person available for the next one.
Say what happened
"You do not match this study" is a better experience than a silent redirect, and it makes the next invitation credible.
Incentives and data integrity
A meaningful incentive attracts effort, and a portion of that effort will go into qualifying rather than answering honestly. Design for it rather than hoping.
Use screener questions whose right answer is not guessable from the question. Asking "do you manage a budget over $50,000?" tells the respondent exactly what to say; asking what they are responsible for, in their own words, does not.
Include a small number of items with a known answer, scored and used to weight or exclude. Include attention checks sparingly — they catch the careless and irritate the careful, and a heavy hand costs you more good responses than bad ones.
And where the study depends on something being true about the person, ask for evidence that is a by-product of it being true rather than an assertion. What counts as evidence and how to specify it before the work starts covers the formats and, importantly, the sequencing: a requirement invented afterwards cannot be applied to responses already collected.
Every screener teaches respondents what you want. The good ones teach them nothing.
Multi-market studies, where one number does not travel
Setting a single incentive across eleven countries is the most common shortcut and it produces uneven samples in a predictable direction.
The same amount represents very different value in different economies. Pitched at a level that works in a high-income market, it can be large enough elsewhere that qualifying becomes worth optimising for; pitched at a level that works in lower-income markets, it fills slowly or not at all in the others, and the markets that under-fill are quietly dropped from the analysis.
Scale the incentive by market rather than converting one figure at the exchange rate, and check what a competing platform pays locally rather than reasoning from your own cost of living.
Payment method matters as much as amount. A gift card for a retailer that does not operate in a country is worth nothing there, and bank transfer availability, minimum thresholds and fees differ enough that a nominally equal incentive can arrive as noticeably less. Ask what actually reaches somebody in each market after fees.
The cases with special rules
Three, and each needs advice rather than a rule of thumb.
Healthcare professionals. Payments to clinicians are reportable in several jurisdictions and many employers have their own policies. This is a legal question with real consequences and the answer differs by country and by role.
Public sector employees and officials. Rules on accepting payment vary widely and some prohibit it outright.
Your own employees, when studying internal products. Paying them is usually unnecessary and occasionally problematic; the more useful step is making participation genuinely voluntary and visibly separated from their manager.
For anything in the first two categories, take advice before designing the incentive rather than after the study fills.
Your own customers are a different case
Surveying people who already have a relationship with you changes the calculation in both directions.
Response rates are higher, so a smaller incentive goes further. But a large incentive to an existing customer can read as compensation for a problem, which colours the answers, and the goodwill of the relationship can produce softer criticism than a stranger would give.
For genuinely candid input about a product people pay for, the more useful population is often the one that left or the one that chose a competitor, and neither will answer for a product credit. Which populations a customer feedback programme never reaches covers that gap and what closing it costs.
What to ask a supplier
Five questions worth putting to any panel or recruitment partner.
What proportion of the incentive reaches the respondent? A partner paying a small fraction will struggle to hold the people you want, and the shortfall shows up as sample quality rather than as a line on the invoice.
How quickly are respondents paid after completion, measured rather than promised?
What is the policy on screen-outs?
How often can one person take part, and how is that enforced? This is the main defence against professional respondents.
And what is the incidence estimate for our criteria, and what happens to the price if it turns out to be wrong? How each layer of targeting narrows the pool and moves the rate sets out why that estimate matters more than the incentive itself.
For the participant's view of the same market — what studies actually pay, and how the screening feels from the other side — what paid research studies are worth is the honest account, and it is worth reading before setting a number somebody else has to find acceptable.
The incentive only matters once the method is right. Which research method actually answers your question covers the step before this one, including the surveys that should have been interviews.
The questionnaire on the other side of the incentive matters just as much, and what to ask and in what order sets out the design faults that cost more than any payment decision.