7 Sites That Paid You in 2024 and Stopped in 2026
A payout that used to land on the same date each month stops landing. The app still opens. The balance still displays, sometimes for weeks. Then the store listing disappears, the support inbox bounces, and the number that used to be money turns into a screenshot you keep for no practical reason.
Every platform below was still paying people in 2024. Their shutdowns run from February 2025 through September 2026, and all seven are offline now. What separates the ones that paid out on the way down from the ones that kept the balance is narrow and predictable: whether the closing notice named a withdrawal deadline, and whether there was still a functioning process behind it.
What follows is each case, with dates.
The four questions that decide whether you get paid
When a rewards or gig platform closes, your balance becomes an unsecured claim. You are one of thousands of creditors, and the practical outcome comes down to four things.
- Notice length. A four-day warning and a three-month warning are different events.
- Whether a deadline was published. “We are closing” and “withdraw by this date” are not the same notice.
- What the balance was denominated in. Fiat currency can be wired. Platform tokens usually cannot be recovered once the servers stop.
- Whether a payout rail was already linked. A verified, still-active bank account or PayPal connection turns a closure into an automatic transfer. Without one, you are filing a manual claim and hoping the inbox is still monitored.
Notice length is the variable people underestimate. Presearch gave users four days. Cashrewards gave them three months, and most members never had to file anything. The second was survivable by accident. The first required you to be reading your email that particular week.
Shopkick: no notice at all, closed 26 March 2026
Shopkick paid points called kicks for walking into stores and scanning products. On 26 March 2026, the app and the website were both gone. There was no announcement to users, and reports at the time said employees were not told either.
Unredeemed kicks had no cash value and no claim process. The terms users had agreed to years earlier contained the line that mattered: that Shopkick was under no obligation to notify users if kicks were no longer available.
Two months later, there was at least a formal answer. Shopkick, Inc. filed for Chapter 7 bankruptcy in the United States Bankruptcy Court for the District of Delaware, Case №26–10877. That docket is the only process that now exists, and a kick’s balance enters it as an unsecured consumer claim ranked behind every secured creditor. In a liquidation, that is worth nothing in practice.
The clause that was always there
That sentence had been in the agreement the whole time. It is the standard construction in loyalty programs, and it means the notice you were waiting for was never owed to you. One user reported roughly 838,000 kicks stranded, which at Shopkick’s baseline rate of 250 kicks to the dollar came to about $3,350, with no mechanism to ask for it.
The worst case is not a bad payout schedule. It is a company that stops existing before it has to explain itself.
Presearch: four days from notice to offline, July 2026
Presearch announced on 24 July 2026 that it would go offline on 28 July at 5 pm PST. The notice told users to withdraw their tokens before that deadline, and added: “Please don’t wait until the last day.” Support requests were accepted until 31 July.
Four days. And the balance was denominated in PRE tokens rather than dollars, which meant withdrawal was not a transfer to a bank account but a move to a wallet the user controlled. Anyone who had never set one up had four days to learn how.
Why token balances are the harshest category
A fiat balance is a debt the operator can still pay, and in most jurisdictions it survives as a claim even after the site goes dark. A token balance is only worth something while the platform’s own infrastructure runs. When that infrastructure is switched off, the tokens do not transfer anywhere or convert into anything. They stop existing in any usable sense, and there is no equivalent of a claims window because there is no counterparty left holding funds.
Amazon Mechanical Turk: an announced close, 25 August to 30 September 2026
MTurk was the opposite of Shopkick. Amazon announced on 25 August 2026 that the platform would permanently close on 30 September 2026, and the notice addressed payment directly. Completed and approved HITs would be paid out during the normal scheduled disbursement cycle. Requesters kept a 30-day window to approve submitted work, and anything not acted on within that window auto-approved. Bonuses could still be issued until 30 October, and transaction history stayed available until 28 January 2027.
That is about as orderly as a shutdown gets. The platform told workers what would be paid, when, and where to find the record.
The gap is the word “approved”
Work that had been submitted but not yet approved by the cutoff sat in limbo, and approval was never something the worker controlled. The 30-day auto-approval backstop meant it would eventually clear, but “eventually” is not the same as “on the normal disbursement cycle,” and a worker who needed the money in October was still waiting on a requester who had already moved on. On a marketplace with tens of thousands of requesters, some fraction of them always disappears.
So the split was not honest versus dishonest. It was approved versus pending, and the pending category had no appeal — only a delayed automatic release.
X Creator Revenue Sharing: three final payouts, August to September 2026
X stopped accepting new enrollments into Creator Revenue Sharing on 7 August 2026 and retired the program on 7 September 2026. Enrolled creators received three final payouts, on 14 August, 28 August, and roughly 11 September.
Two mechanics decided who actually collected. The first was the $30 minimum payout threshold, which had to be cleared before a transfer would trigger. The second was eligibility, since the program required 5 million impressions in a three-month window and 500 verified followers. A creator whose reach dipped in the final quarter could stay enrolled and still earn nothing.
The threshold trap
A minimum payout is harmless while a program runs indefinitely. On a program with a published end date, it becomes a deadline you can miss without noticing, because nothing about the balance display tells you it is about to become uncollectable. Anyone sitting at $22 in early September had no path to the $30 that triggered the transfer.
Get Reham’s stories in your inbox
Join Medium for free to get updates from this writer.
X replaced the program with an application-based Original Content Rewards scheme. That is a different arrangement with different rules, not a continuation of the old one, and enrollment is not automatic.
Cashrewards: a three-month claim window, September to December 2025
Cashrewards, an Australian cashback site with roughly two million members, stopped offering new deals at midday on 8 September 2025, after eleven years of operation.
Members who already had a current bank account or PayPal connection linked had their balances transferred automatically. Everyone else had to act. Balances had to be withdrawn by 24 October 2025, and manual claims ran until 12 December 2025. Gift cards that had not been downloaded before the site was decommissioned were gone.
The setting that decided it
Two members with identical balances on the same day got opposite outcomes based on a payout preference most of them had configured once, years earlier, and never looked at again. One received a transfer. The other had to notice an announcement, find a claims form, and complete it inside a two-month window.
This is the case that makes the argument for keeping payout details current, and it is the reason that item sits first in the checklist below.
Dosh: a six-week exit window, January to February 2025
Dosh announced its closure in January 2025. The last day to make a qualifying transaction was 16 January. The app shut down on 28 February, and withdrawals had to be requested by 23 February. Balances not withdrawn by that date were forfeited.
Notice the spacing. The earning cut-off came six weeks before the app closed. Users who kept opening the app through February saw a balance that looked like it was still accumulating, when nothing they did after 16 January counted toward it.
A closure notice containing two different dates, one for earning and one for withdrawing, is the easiest kind to misread, and the gap between them is where the money goes.
Kindle Vella: the dashboard disappeared, the payments did not, February 2025
Kindle Vella closed on 26 February 2025. Episodes stopped being purchasable, the author dashboard was retired, and reporting on episode performance went away. What did not go away was the royalty stream. Amazon’s guidance to authors was that payments would continue to appear in the Payments statement as usual.
This is the least dramatic outcome of the seven, and it is the one people misjudge in the other direction. When the interface disappears, the instinct is that the money has too.
When inaction is correct
Here, doing nothing was the right move. There was no claim to file and no deadline to miss, and anyone who panicked and deleted their KDP tax or payment records made their own situation worse. The skill in a shutdown is telling the two cases apart, because the correct response in one is the exact opposite of the correct response in the other.
What actually decides the outcome
Line the seven up, and the pattern is not about company size or how long a platform had been running. MTurk had 21 years behind it and still left a pending-work gap. Shopkick had a large user base and no process at all.
The variables that predicted the result were administrative:
- How much notice users received, measured in days rather than in principle
- Whether the notice contained a date, a process, and a working contact
- Whether the balance was fiat or token
- Whether a payout rail was already verified and current
The pattern I keep running into is that the financial question and the practical question come apart. A platform can owe you money in a way that is legally clear and still leave you with nothing, because the amount was below a threshold, or denominated in a token, or waiting on an approval that never came. Solvency mattered less than paperwork.
The same structure appears outside shutdowns. Fiverr’s 90-day safety period is a fixed number, but the release is still subject to Fiverr’s approval, which means the calendar can pass without the money moving. A deadline and a condition are not the same thing.
What to do before the next one closes
None of this requires predicting which platform dies. It requires removing the parts of the outcome you control.
- Keep a payout rail linked and current. A verified bank account or PayPal connection is what turned the Cashrewards wind-down into an automatic transfer for most members. A card that expired in 2023 is not a payout rail.
- Withdraw at the threshold, not at a target. If a platform pays out at $25 or $30, take the transfer when you cross it. Sitting on a balance to let it grow is a bet that the program outlives your patience.
- Export transaction history every quarter. MTurk kept records until January 2027. Shopkick kept none. You do not know in advance which kind of operator you are dealing with.
- Read the expiry clause once. Search the terms for expire, forfeit, and no obligation to notify. If any of those appear next to the word balance, treat the number on the dashboard as provisional.
- Watch the announcement channel, not the app. Presearch’s notice went out by email and newsletter. Users who only opened the app found out after the servers were off.
Conclusion
Seven platforms, seven different endings, and the same underlying structure. Your balance is a claim against a company, and the strength of that claim depends on terms you agreed to without reading and settings you configured once and forgot.
The shutdowns that went well for users were not the ones run by the biggest companies. They were the ones where the operator stayed long enough to publish a date and process a claim, and where the user had already given the platform somewhere to send the money.
That is a low bar, and it sits entirely on your side of the line.
