Paying yourself to do your own tasks: what the incentive trials show, and whether an earn-back subscription could work

A delivery app can get someone off the sofa and across a city for a modest fee, while their own short list of small jobs sits untouched. The obvious response is to build the dispatcher for yourself and put money behind the orders. The research on that idea separates two ways of doing it, and treats them very differently.

1What a delivery order bundles that a to-do list does not

The comparison is more exact than it first sounds. The research breaks a delivery order into its parts: A DoorDash order bundles at least six things: an external queue that decides what's next; one order at a time; a concrete, bounded action with an ETA; a trigger (the ping); immediate feedback; real money that a third party pays you and that you need.Key findings Most of that list is design, not money: The first five are all things you can build.Key findings The last is the hard part: The sixth is different in kind.Key findings

2Paying people does work, while you are paying

The trials are consistent on the basic effect. A meta-analysis of 34 randomised trials found incentives moved behaviour with the odds ratio was 1.53 up to 18 monthsTL;DR, though effects mostly faded within about 3 months of removalTL;DR. A later review describes them the same way — modest, often short-livedKey findings. One design detail beats the rest: framing the money as something you already hold and could lose. only the loss arm significantly beat controlKey findings.

The broader evidence is mixed in instructive ways. Rewards can undermine motivation, but mainly for work people already enjoy: Rewards crowd out intrinsic motivation — strong for interesting tasks, weak for boring ones (Deci, Koestner & Ryan 1999).Key findings Paying very little can be worse than paying nothing: Small pay can backfire — moderate (Gneezy & Rustichini 2000).Key findings And honesty slips when people grade themselves: Self-verification invites small cheating — moderate to strong (Mazar, Amir & Ariely 2008; the 2018 RRR did not replicate the moral-reminder effect).Key findings Pairing a chore with a treat helps for a while: Temptation bundling helps — moderate, decays (Milkman 2014).Key findings

3Paying yourself a wage loses most of it

The weak version is the one where you fund a wage out of your own pocket and pay it to yourself per task. The money is not income — Mental accounting predicts people treat that money as "my money moving between pockets", not as earnings.Key findings Nor does anyone check the work. The research states the loss in one line: Self-verified pennies remove all three levers that work in the trials: outside money, objective verification, meaningful amounts.Key findings But it names the condition under which your own money can still carry weight: That is why the money layer needs either loss framing (you lose something you already hold) or outside verification to carry any weight.Key findings

4Earning a subscription back is a different mechanism

A subscription you earn back by doing the work meets that condition. The fee has already left your pocket; doing the tasks recovers some of it, and skipping them leaves it with someone else. That is a stake, not a wage — and the research's own conclusion is that it is the strongest version of the idea: The version most likely to work is a dispatcher that rewards you with loss-framed credits you earn back against your own subscription.Summary The design it sketches is deliberately modest: earn-back subscription credit (loss-framed, bounded, no money leaves the app)Details.

Where it draws a line is at the step of pooling what other people lose and paying it out to whoever finishes: Avoid user-funded forfeit pools: they collide with Google Play's real-money staking ban, Canada's payment-services rules and the Criminal Code's lottery provisions.TL;DR That is a legal constraint on one way of distributing the money, not evidence that the incentive fails. The research's verdict on the business is conditional rather than closed — it probably works as a subscription dispatcher with rebate credits; likely fails as a cash-payout platform without a B2B payer.Synthesis

Paying out small sums in real money also runs into arithmetic. Payment processors charge per transfer, and at the scale of a few cents per task the fees swamp the payment: Paying a 5¢ commission would cost roughly 5–12× the commission.Key findings A credit against a fee already paid avoids that cost entirely, which is part of why the research prefers it.

5Verification is still the hinge

Whichever way the money moves, someone has to decide a task was done. The clearest cautionary case is a commitment app that charged people for sessions they had actually completed. It shut down in July 2017, and an FTC settlement required it to refund $948,788 to users it had charged even when they met their goalsTL;DR. The diagnosis is not that the idea was wrong — Its problems were verification and billing, not the core idea.TL;DR Who referees matters at the margin: self-refereed commitments succeeded 68.7% success self-referee vs 74.2% friend-refereeKey findings. And the appetite for real stakes is small — 32% of the commitment contracts included stakesKey findings.

6What has been tried, and the part nobody has built

The neighbouring products cover pieces of the idea. Commitment apps take a stake and keep it if you fail; Beeminder was, in 2017, still tiny ($22k/mo revenue with 3.5 FTEs)Key findings. A newer one reviews proof with AI and people and claims 94% completion, unauditedKey findings. Calendar schedulers dispatch work too, but by time rather than place: Motion, Reclaim, Sunsama (dispatch by calendar time, not context)Key findings.

What none of them does is the delivery app's core loop. The research found no product that turns brain-dumps into single, context-triggered "orders", dispatches them one at a time from a quota, requires a reason when you decline, and re-prices the work dynamically.Key findings It singles out one by-product as the most valuable: Of all the features, the decline-reason data is probably the most valuable and least copied.Key findings The failures it catalogues share a pattern: verification is the business; very few users opt into losing money (32% of stickK contracts; 13.7% in Halpern et al.); micro-cash is eaten by fees; "earn-money" apps carry a store-policy stigma.Key findings

7For ADHD, frequency beats size

The ADHD literature supports the dispatch rhythm more than the payment. Reinforcement helps — reinforcement contingencies have a positive impact on task performance and levels of motivation for both children with AD/HD and normal controlsKey findings — and the responsiveness is to how often, not how much: sensitive to frequency but blind to the magnitude of penaltyKey findings. Many small immediate signals do work that one large delayed payout does not, which suits a queue of small credits better than a single refund at the end of the month.

8None of it has been tested directly

For a question this specific, the direct trial does not exist. No RCT tests cash micro-incentives for task completion in adults with ADHD — a genuine evidence gap.Key findings The nearest academic analogue tested whether varying the payment — surge pricing for your own chores — beats a flat amount, and found it does not: no significant benefit, 0.66 success fixed vs 0.61 adaptive (P=.54); incentive size no significant effect (OR 0.98)Key findings. The earn-back version has no trial at all. The research's reading of the field is that the gap is real rather than crowded: So the combination is genuinely open space.Summary

9What this explains

It explains why the delivery-app comparison feels so sharp and transfers unevenly. Five of the six things that make an order easy to accept — an external queue, one job at a time, a bounded task, a ping, immediate feedback — are structure, and structure can be copied. The sixth is money that matters. Coins moved between your own pockets do not supply it. A fee you stand to lose might, and the research's best guess is that a subscription you earn back carries the weight a self-paid wage does not. That guess is untested, and it still depends on the hinge that broke the last company to try: a credit is only a stake if claiming it takes more than a tap. The idea is not closed. It is narrowed to a version that can be tried, and the trial has not been run.