Written by Krisztián Stark
Key takeaways
- Reversal rate is the share of credited conversions, or of credited payout, that was later taken back, measured for a set period or a cohort of credits.
- Most reversals come from fraud checks, VPN or duplicate accounts, unmet requirements found later, and refunds, and each cause leaves a different signal in your data.
- There is no universal good number: compare each source, country, and offer with your own baseline, and act on spikes and outliers rather than on the average.
What is a reversal rate?
A reversal rate is the share of credited conversions that were later taken back. On an offerwall, a user completes an offer, the result is credited, and the publisher pays the user a reward. If the advertiser later rejects that result, for example after a fraud check or a refund, the credit is reversed and the payout comes back out of the publisher’s balance. You can measure the rate by count (how many conversions were reversed) or by value (how much payout was reversed), and both matter.
In affiliate marketing the same event is often called a chargeback.
How to calculate your reversal rate
Use two formulas side by side:
Reversal rate by count = reversed conversions / credited conversions × 100
Reversal rate by value = reversed payout / credited payout × 100The count rate tells you how often something goes wrong. The value rate tells you how much it costs. They often differ, because fraud tends to aim at high-paying offers: a few reversals on large offers can cost more than many reversals on small tasks.
Period view vs cohort view
The next choice is which reversals you count against which credits. There are two honest ways to do it:
- Period view. Reversals that arrived this month divided by credits that arrived this month. It matches the money that left your balance, but mixes reversals of older credits with this month’s volume.
- Cohort view. Take all credits from one period (a week or a month), then count how many of those exact transactions were reversed, whenever the reversal arrived. This shows the true quality of the traffic you sent in that period.
Card payments have the same split. Stripe’s guide to measuring disputes separates dispute activity, counted by dispute date, from dispute rate, counted by payment date, and notes that the dispute rate for recent payments can still change because cardholders can dispute up to 120 days later. Your cohort rate behaves the same way: a cohort from last week looks cleaner than it really is, because some of its reversals have not arrived yet. Compare cohorts of the same age, not a fresh cohort with an old one.
Illustrative example: sample figures only
A rewards site looks at its September cohort on October 21. It credited 1,200 conversions worth $1,500 in payout. Since then, 30 of those transactions were reversed, worth $90. The reversal rate is 2.5% by count (30 / 1,200) and 6.0% by value ($90 / $1,500). These numbers are invented to show the math. They are not a benchmark for any site or offer.
Why offer rewards get reversed
Advertiser fraud checks that run after the credit
Many advertisers measure installs and in-app events with a mobile measurement partner (MMP), and some of those checks only finish after the result was reported. AppsFlyer’s Protect360 anti-fraud guide says post-attribution fraud can be identified on the day of the install and up to 7 days after (checked on October 21, 2026). When an advertiser’s tool flags a result that was already credited, the offer step is reversed.
VPN, proxy, and data center traffic
Advertisers pay for real users in the countries they target. A user who hides their location behind a VPN to reach offers from another country is not the user the advertiser paid for. Adjust’s anonymous IP filtering page describes checking traffic against a database of anonymizing services, such as a VPN, a Tor exit node, or a known data center, and excluding post-install activity from those services from reporting. Its rejected install callbacks list reasons such as anonymous_traffic, click spamming, and click injection.
Multiple accounts and duplicate users
Many offers pay once per person or once per device. When one person runs several accounts on your site, or reinstalls an app to repeat an offer, the advertiser sees the same device or the same identity again and rejects the repeats. Stopping multi-accounting and VPN abuse covers how this looks in your own data.
Requirements that were not met after all
On Sharklio, a step the advertiser does not confirm is never credited in the first place, so most unmet requirements show up as missing rewards, not reversals. A reversal happens when the advertiser confirmed a step and later found it did not count, for example a trial that was cancelled within a qualifying period, a sign-up with fake details, or a level reached by breaking the game’s rules. Why users start offers but don’t finish explains how unclear requirements cause problems on both ends.
Refunds and payment chargebacks
Offers that involve a purchase or a paid subscription can be reversed when the user gets their money back. Apple’s refund request page says some App Store purchases might be eligible for a refund, and Stripe’s disputes overview explains that a card dispute, also called a chargeback, reverses the payment and takes the funds from the business. If the advertiser loses the sale, it usually stops paying for the conversion too.
A credited result that broke the offer rules
Sharklio’s Credits and reversals docs list one more reason: staff found that a credited result broke the offer rules. Rewarding users for rating apps is one example of a rule break. The traffic quality rules for publishers list what is allowed.
How a reversal reaches you as a publisher
On Sharklio, a reversal never edits or deletes the postback you already received. It arrives as a new, signed server-to-server postback for the same transaction:
| Macro | Credit | Reversal |
|---|---|---|
{transaction_id} | New ID for this result | Same ID as the credit |
{status} / {state} | 1 / credited | 2 / reversed |
{reward} and {payout} | Positive amounts | The same positive amounts, never negative |
{event_id} | Unique | New and unique |
Your server takes {reward} back from the user in {user_id}, but only if it credited that transaction_id, and only once. It answers with HTTP 200 even when it ignores the event, so the call is not retried. A reversal is never sent before the credit it cancels. The Postbacks reference has a complete handler, and What is a postback URL? explains statuses and macros in plain terms. Always verify the signature before you act on either status.
On the Sharklio side, the payout is deducted from your balance, the user sees the offer as Reversed in My offers on your offerwall, and the result is listed under Reports, Chargebacks with the date and the reason if one was given. Publisher earnings and reversals covers what that means for your balance.
Store the context with every credit
To compute your own reversal rate by segment, save these macros on the credit row: {offer_id}, {country}, {device}, {integration}, and {sub1} to {sub5}. When a status 2 postback arrives, match it to the row by {transaction_id} and mark it reversed with the date. Every segment report below then becomes one query.
Reversal causes, signals, and fixes
| Cause | Signal in your data | Fix |
|---|---|---|
| Multiple accounts | Reversals cluster on a few users who share an IP, device, or payout account, often all new | One account per person in your terms, sign-up checks, review accounts that share signals |
| VPN or location spoofing | Reversals concentrated in high-paying countries from users whose sign-up country differs | Keep VPN blocking on, compare sign-up country with offer country, ban repeat offenders |
| Bots or scripted completions | Completions faster than a person could manage, at odd hours, in bursts | Rate limits, human checks at sign-up, stop accounts that complete offers too fast |
| Requirements not met | Reversals spread across many normal users on one offer or offer type | Make requirements clear on your earn page, stop pushing that offer in promotions |
| Refunds and payment chargebacks | Reversals on purchase, trial, or subscription offers, arriving weeks after the credit | Longer holds on those offer types, remind users what cancelling a trial means |
| Low-quality traffic source | One sub ID, campaign, or partner has a far higher rate than the rest | Pause or cap that source, ask what it promised users |
| Rule breaks on your side | Reversals with a reason that names the offer rules | Remove incentives the offer does not allow, such as rewards for ratings |
For deeper fraud patterns such as click spamming or emulators, see ad fraud and bot traffic.
What a healthy reversal rate looks like
There is no published industry number for a healthy offerwall reversal rate that would fit your site. It depends on your offer mix (purchase and subscription offers reverse more than short tasks), your countries, your users, and how long you wait before measuring. Your own baseline is the number that matters.
Set your own baseline
- Pick mature cohorts. Use monthly cohorts that are old enough for most of their reversals to have arrived. Record the days between each credit and its reversal to see how long that takes for your offer mix.
- Split it. Compute the rate by count and by value for each traffic source (sub ID), country, offer, device, and new vs established users.
- Note the normal range. After a few cohorts, each segment has a range it usually stays in. That range is your baseline.
- Review on a schedule. Compare every new mature cohort with the baseline, and look again whenever you add a traffic source or a promotion.
Small segments are noisy: five reversals out of twenty credits may be one bad user. Look at the users behind the reversals, not only the percentage.
Signs that a reversal rate is worrying
- A spike. A cohort well above the range of the ones before it, especially after a new promotion or source.
- Concentration. Most reversals in one sub ID, one country, one offer, or a handful of users.
- Value far above count. Reversals hitting your highest-paying offers, which is where fraud aims.
- New accounts. Reversals that come mostly from accounts created in the last few days.
Illustrative example continued: sample figures only
Split by {sub1}, the same September cohort shows where the problem is. Organic site traffic: 700 credits, 4 reversed (0.6% by count, 2.5% by value). Discord community: 300 credits, 3 reversed (1.0% and 3.75%). A giveaway post on a coupon forum: 200 credits, 23 reversed (11.5% and 18.3%). The blended 2.5% hid one source responsible for most of the losses.
Click IDs and sub IDs explains how to tag sources so this split is possible, and how to read your dashboard and reports shows where the Conversions and Chargebacks lists live.
How to lower your reversal rate
- Hold new rewards. A hold does not stop reversals, but it makes them free: the reward is cancelled before the user can cash it out. Tier the hold by account age, amount, and offer type, as described in reward hold periods.
- One account per person. State it in your terms and enforce it with checks on shared devices, IPs, and payout accounts. Device fingerprinting is one of the signals.
- Block VPN and proxy traffic. Keep users on their real location so they only see offers they can qualify for.
- Make requirements clear. Tell users to read the offer steps, finish them on the same device, and not cancel a trial early if the offer needs it.
- Analyze by sub ID. Tag every traffic source, then cut or cap the ones whose reversal rate sits far above your baseline.
- Watch completion speed. Stop accounts that complete offers faster than a person could.
What not to do
Do not reward users for rating apps or for offers they did not complete, do not promise rewards for steps the offer does not pay for, and do not reverse rewards on your own without a status 2 postback. Each of these either raises your reversal rate or creates support tickets you cannot answer.
How to handle reversals with your users
Take the reward back, once
When a signed status 2 postback arrives for a transaction you credited, subtract the same {reward} from that user. If the reward is still on hold, cancel it and the user loses nothing they could spend.
Decide your negative balance policy in advance
If the user already spent the reward, you have two common options: let the balance go below zero so future earnings pay it back, or hold withdrawals until it is recovered. Either way, write it into your terms before users earn, and apply it the same way to everyone.
Support wording that calms instead of accuses
Most users who see a reversal did not commit fraud. Explain what happened without blame. A reply you can adapt:
Example reply for a reversed reward
“Thanks for asking. The advertiser reversed this offer after it was credited, so the 450 points were removed from your balance. This usually happens when an offer’s requirements were not fully met, the purchase or trial was cancelled, or the advertiser’s checks flagged the completion. You can see the status in My offers on the offerwall. If you believe this is a mistake, reply with the offer name and the date you completed it and we will look into it.”
Handling missing reward tickets has more templates, including how to explain a reversal without starting a fight and when to escalate.
How Sharklio reports reversals
Sharklio is a pay-per-result affiliate network. Every reversal reaches your server as a signed status 2 postback with the same transaction ID and the same positive amounts as the credit, so your ledger can match and take back the exact reward. Your dashboard and reports show clicks, conversions, and reversals by country, offer, and sub ID, and a Reporting API brings your numbers into your own tools. Users on VPNs, proxies, Tor, or data center networks are blocked from the wall. See how the wall works or get in touch.
Frequently asked questions
What does “offer reversed” mean?
It means an offer that was already credited was taken back by the advertiser, for example after a fraud check, a refund, or finding that a requirement was not met. The reward is removed from the user’s balance and the payout from the publisher’s.
Why was my offer reward reversed?
The most common reasons are using a VPN, having more than one account, not completing every requirement, cancelling a purchase or trial, or the advertiser’s fraud checks flagging the completion. On a Sharklio wall, the offer shows as Reversed in My offers.
Is a reversal the same as a chargeback?
In affiliate marketing the two words are usually used for the same thing: a conversion that was credited and later taken back. A card chargeback is a related event, where a cardholder disputes a payment with their bank.
How do you calculate reversal rate?
Divide reversed conversions by credited conversions for the same period or cohort and multiply by 100. Do the same with payout amounts to get the rate by value, which shows the cost.
What is a good reversal rate for an offerwall?
There is no single number that fits every site, because offer mix, countries, and users differ. Build your own baseline from mature cohorts per source, country, and offer, and investigate any segment that rises well above it.
Can a reversed offer be credited again?
On Sharklio, a transaction is credited once and can later be reversed, and nothing else is possible, so a reversed transaction is not credited again. If a user believes a result was reversed by mistake, send the user ID, offer name, and date to Sharklio so the team can check it with the advertiser.
Should I take the reward back from the user?
Yes, when you credited that transaction, and only once. Holding new rewards before cash-out keeps this from costing you or the user anything they already spent.
