How to automate survey incentives, and why that matters

Ask a room full of research and insights people how they pay participants and you'll get multiple answers phrased about the same way: we send rewards to participants. Sometimes it's a Visa card. Sometimes it's Amazon. Sometimes the panel provider handles it and the cost is buried in a line item nobody reads. And when there is no panel provider, someone on the team opens a spreadsheet, sorts by completion status, and starts copying email addresses into a gift card portal.

That process is not wrong. It's just expensive in a way that doesn't show up on any invoice.

The spreadsheet is doing four jobs at once

The thing people call "the tracking sheet" is quietly acting as four separate systems, none of which it is good at.

The participant roster. Who was invited, who screened in, who screened out, who is mid-survey, who abandoned at question 14. This changes constantly during fielding, and the sheet only reflects reality as of the last time somebody exported the survey platform's response file.

The eligibility record. Did this person actually complete, or did they hit the final page and bounce? Did they pass the attention check? Did they already take this study last quarter under a different email? A spreadsheet cannot answer the last one without a lookup nobody has time to build.

The fulfillment queue. Highlighted rows mean "paid." Uncolored rows mean "not yet." Nothing enforces this. If two coordinators work the list on the same afternoon, somebody gets paid twice and somebody gets skipped.

The audit trail. When finance asks what the study cost, or when a participant emails three weeks later saying they never got anything, the sheet is the only evidence that exists. It has no timestamps, no delivery receipts, and no record of the card that bounced.

Any one of these is manageable. All four in one file, edited by three people, is where the week goes.

What does manual incentive tracking actually cost?

Manual participant fulfillment does not fail randomly. It fails in the same five ways every time.

Typos become unpaid participants

A mistyped email address is not an anomaly, it is a rate. Somewhere between one and three percent of self-reported email addresses are wrong — a transposed character, a .con instead of a .com, an autocorrected name. At 200 participants that is a handful of people you can chase down personally. At 8,000 it is 160 people who did the work, got nothing, and have no idea whether they were ignored or defrauded.

Nobody notices until they complain, and most don't — they just decline the next invitation.

Double-sends and missed sends

Manual queues have no locking. The single most common fulfillment error is not a missing reward, it is a duplicate one, because the sheet was open in two tabs and the second one overwrote the first one's highlighting. Duplicates cost money quietly; misses cost trust loudly.

Reconciliation takes longer than fulfillment

Sending 500 gift cards takes an afternoon. Proving you sent 500 gift cards, to the right 500 people, for the right amounts, takes considerably longer — because the evidence lives in three places that disagree. The survey export says who completed. The gift card vendor's order confirmations say who was issued a code. The credit card statement is the only source with real amounts, and it shows one lump charge.

Reconciling those three is archaeology, and it happens at the exact moment the team wants to be analyzing data.

Raffles are a workaround, not a strategy

There is a very common trick for stretching a fixed incentive budget: instead of paying everyone, enter everyone into a drawing for a few larger prizes. Five $100 cards instead of 500 $10 cards. It genuinely works — it is cheaper to administer, and a large prize can be more motivating per dollar than a small one.

But run manually, it introduces its own problems. Who is actually in the pool? How was the winner drawn — a random number generator, or somebody scrolling and stopping? Can the panel see that a draw happened at all? A raffle whose winner is never announced is indistinguishable, from the participant's side, from a raffle that never happened. That is the version they will assume.

Nothing is instant

The gap between completion and reward is where goodwill dies. A participant who finishes a twenty-minute survey and gets a card in their inbox before they close the tab will take your next study. A participant who is told "rewards go out at the end of the month" has been given a promise from a stranger, and a meaningful percentage of them have been burned by that promise before.

Manual fulfillment is batched by definition, because no one is sitting there sending cards one at a time in real time. Batching is the thing that makes rewards feel slow.

What automation actually needs to do

"Send the emails faster" is not the requirement. The list is more specific:

Track participant state continuously, not at export time
Verify the completion event before releasing anything of value
Deduplicate against prior studies and against itself
Deliver instantly on qualification, not on a monthly batch
Record a per-participant receipt with a timestamp and a status
Retry bounces automatically and escalate the ones that keep failing
Enforce a hard budget ceiling nothing can cross
Make raffle mechanics visible and provable to the people entering them

Only item seven is really about money. The rest are about never again having to answer "did participant 4,812 get paid?" by opening a file.

How Sentiv handles it

Sentiv Rewards is built around exactly this problem: the campaign, the participants, and the reward delivery live in one system, so the tracking sheet stops being necessary.

Participant tracking and reporting

Every participant in a Sentiv campaign has a record with a live status — invited, started, screened out, completed, rewarded, failed — updated by the events themselves rather than by a periodic export. Because that record is per-participant, the questions that used to require archaeology are lookups:

Who completed but has not been rewarded yet?
Which deliveries bounced, and what happened on retry?
Has this email address appeared in any prior campaign?
What is this study's spend, right now, against its budget?

Reports run off the same records, so the fulfillment report and the finance number are the same number. No reconciliation step, because there is nothing to reconcile.

Instant rewards

An instant reward campaign fires on a verified completion event. Participant finishes, qualification is checked against the rules you set, and the gift is delivered — typically while the person still has the tab open.

Rules are configurable rather than manual: one reward per person per campaign, a minimum time in survey to filter speeders, an attention-check requirement, an exclusion list from earlier studies. Everything the highlighted-row system was approximating, enforced automatically.

Raffles and sweepstakes

For teams stretching a fixed budget, Sentiv runs raffle campaigns as a first-class type rather than a manual workaround. You define the prize count and value, the entry window, and the qualification rule. Entries accrue automatically as participants qualify — no separate list, no copying names into a randomizer.

The draw is executed by the platform, logged, and the winners are notified and fulfilled through the same delivery pipeline as an instant reward. Participants can see the entry count and the stated odds, which is the part that makes a raffle believable. A drawing with "1 in 240" printed on it converts better than one that says "enter to win," because the number is evidence the drawing exists.

Automatic gift delivery

Delivery is handled end to end. Sentiv issues the gift from the catalog you have configured, sends it to the participant, and records a receipt with a timestamp and a delivery status. Bounces are detected and retried rather than discovered six weeks later, and persistent failures surface in a queue you can act on instead of disappearing.

Budget ceilings are enforced at the campaign level. When a campaign hits its cap, delivery stops — a runaway integration cannot spend money you did not allocate.

When should you stop using a spreadsheet?

If you are running one study a quarter with 150 participants, the spreadsheet is fine. It is genuinely fine, and the setup cost of any tool is not worth it.

The line is somewhere around the second concurrent study, or the first time a participant emails to say they never got paid and you cannot prove otherwise. At that point the sheet has stopped being a tool and started being a liability — and the hours it consumes are hours that were supposed to go into the research.

Ready to stop tracking by hand? Start a campaign with Sentiv and see what your next study looks like without the spreadsheet.