How to Set a CPA Payout for an Offerwall Campaign

How do you set a CPA payout for an offerwall campaign?

Set a CPA payout by working backward from value. Estimate what a user who completes the action is worth to you, subtract the margin you want to keep, and the result is your maximum CPA. Then start below that ceiling, split it across steps and countries, and raise it only when the results show you have room. On an offerwall, also check that the payout is high enough to show users a reward worth their effort, because part of what you pay becomes that reward.

This guide is the pricing step of a campaign. For the full setup (goal, creative, tracking), see how to advertise on offerwalls, and for the metric itself, what is CPA.

Step 1: know what one action is worth

Every sensible payout starts with one number: the value of a user at the moment they complete the action you pay for. That is not the same as the value of a paying customer, because not every user who installs, signs up, or starts a trial goes on to pay.

Calculate it like this:

• Value per customer. Use lifetime value (LTV) in gross profit, not revenue: payment fees, refunds, and product costs are not yours to spend on acquisition. CAC vs CPA vs LTV shows how to estimate it, and the LTV calculator does the arithmetic.
• Share of users who become customers. Of the users who complete your event, how many pay? Take it from your own cohorts, ideally from users who arrived through a channel similar to rewarded traffic.
• Value at the event = value per customer × that share. If a paying customer brings $40 of gross profit and 30% of users who complete the event pay, a completed event is worth $12 to you.

Use a time frame you can defend. Lifetime is easy to overstate in a young product, so many advertisers price on 90 or 180 days of profit instead and treat anything later as a bonus. If you use model predictions, know what they cover: Google Analytics defines its predicted revenue metric as revenue expected from a user in the next 28 days, and it only becomes available once a property has at least 1,000 returning users who triggered the relevant condition, and 1,000 who did not, over a seven-day period (checked on October 21, 2026). A 28-day prediction is a floor for value, not a lifetime figure.

Step 2: work back to a maximum CPA

Your maximum CPA is the most you can pay per completed event and still hit your profit target. There are two common ways to express it, and they give the same answer:

The break-even line is a limit, not a goal. Break-even ROAS explains why you need a margin above it: your value estimate is uncertain, some conversions will be reversed, and the campaign also has costs that are not in the payout, such as the time spent building tracking. The cost per result calculator turns your value, margin, and close rate into a ceiling in a minute.

Write the maximum CPA down before you talk to any network. Everything that follows is about finding a payout below it that still gets enough completions.

Step 3: check what the user will see

This is the part that is different on an offerwall. Your payout does not stay in one place. The network keeps a share, the publisher receives the rest, and the publisher converts its share into its own currency (coins, gems, points) and passes a part of it to the user. The number next to your offer on the wall is that final amount, shown in the publisher’s currency at the publisher’s exchange rate.

Three consequences follow:

• The reward scales with your payout. The shares and exchange rates differ between walls, but on any given wall, an offer that pays half as much shows roughly half the reward. You cannot pay little and still look generous.
• Users compare effort with reward. They see your offer next to many others and decide in seconds whether the reward is worth the minutes it will take. A long sign-up form or a deep game milestone needs a payout that shows a reward to match. Why users start offers but don’t finish covers this from the publisher’s side.
• Too low is not the same as cheap. A payout far below what similar offers pay usually means few starts and very few completions, which looks like a weak channel but is a pricing problem. Search ads work the same way: Google’s Target CPA help page warns that setting a target too low can reduce conversions (checked on October 21, 2026).

The fix is not to overpay. It is to make the effort match the reward: shorten the action, move the paid event earlier, or split it into steps so the first reward arrives quickly. Incentivized traffic explained covers how rewarded users differ from organic ones and why a clear, honest action matters more than a big number.

Step 4: split the payout across steps

Many offerwall offers pay for several events instead of one: install and open, then a sign-up or tutorial, then a deeper event such as a trial, a level, or a first purchase. Each step has its own payout. Multi-event offers covers which events to choose and how to track them; here is how to price them.

Price the ladder against the deep event

Users who stop after the first step still cost you that step. So the real cost of one user who reaches your valuable event is:

Cost per valuable user = (each step’s payout × users who reach that step, added up) ÷ users who reach the valuable event.

That number, not the last step’s payout alone, has to stay under your maximum CPA. To estimate it before launch, you need the share of users who move from each step to the next. Take it from your own funnel, and expect rewarded users to start more often and drop off differently than organic users, so treat the first estimate as a guess to correct with real data.

Rules for a good ladder

• Keep the first step small and quick. Its job is to get users started and prove the tracking works.
• Put most of the money on the event that predicts value. A generous first step attracts users who collect it and leave.
• Never pay for ratings or reviews. Google Play’s User Ratings, Reviews, and Installs policy says not to offer rewards or incentives for ratings or reviews (checked on October 21, 2026).
• Give every step its own event in your tracking. Adjust’s Add events help page, for example, creates a unique event token for each event, which can be recorded through the SDK or server to server. One token per step keeps the postbacks clean.

If you are still deciding whether to pay for the install at all, CPI vs CPA for apps compares the two models.

Worked example: from user value to a step split

A meal-planning app pays for two steps: create an account (step 1) and start the free trial (step 2). Its data says a trial is worth $12 of 180-day gross profit in Market A and $5 in Market B, and it wants to keep a 50% margin. From past tests it expects about one in four users who create an account to start a trial.

How to read it:

• In Market A, 1,000 accounts and 250 trials cost $500 + $1,000 = $1,500, or $6.00 per trial: exactly the ceiling. That leaves no room for error, so a careful launch would start a little lower, for example $0.40 and $3.50 (4 × $0.40 + $3.50 = $5.10 per trial).
• If only one in five account holders starts a trial, the same Market A payouts cost 5 × $0.50 + $4.00 = $6.50 per trial, over the ceiling. Funnel drop-off changes the price of the deep step, which is why you check the ladder with real reach data after launch.
• Market B gets lower payouts because its trials are worth less to this app. If that shows users too small a reward to start, Market B may not be worth running yet.

Step 5: price each country on its own

A single worldwide payout almost always overpays somewhere and underpays somewhere else. Value per user differs by country because prices, purchasing power, subscription uptake, and refund rates differ. Repeat Steps 1 and 2 per country, or per group of similar countries, and set a separate payout for each.

• Start with the countries where you have real data. How to choose countries for your campaign walks through availability, language, and value per user.
• Check that the reward still looks reasonable to users in each country. A payout that is fine for your value may be well below what other offers on the wall pay there, or well above.
• Use published price data only as a sense check. Cost per install by country shows what public sources report, and tier 1, 2 and 3 countries explains the usual groupings.

Step 6: test, then raise payouts carefully

Your first payout is a hypothesis. Run it long enough to learn, then adjust one thing at a time.

1. Launch below the ceiling. Leave room between your launch payout and your maximum CPA, so you can raise it without losing money.
2. Wait for the full window. Judge a period only after its slowest step has had time to complete and your postbacks have arrived. Deep steps make every campaign look expensive in its first days.
3. Read volume and quality separately. Few starts or completions point to the reward, the effort, or the offer text. Many completions that do not retain or pay point to the event you chose, and a higher payout will not fix that.
4. Raise in steps. If volume is too low and quality is good, raise the payout by a set amount, wait another full window, and compare. Keep each change small enough that you can tell what caused the result.
5. Judge on cost per valuable user. Track spend divided by users who are still active or have paid after 7, 30, or 90 days, by country and by step, not completions alone. A/B testing on a small budget covers how to read small samples without fooling yourself.

Caps, reversals, and confirmation windows

Three settings protect your payout from surprises. Agree on them at the same time as the price.

Caps

A daily, monthly, or total cap limits how many conversions you pay for in a period. It turns your payout into a predictable budget: payout × cap is the most a day can cost you. If the cap fills very early every day, you either have room to lower the payout or can afford more volume. What is an offer cap? explains the types and how resets work.

Confirmation windows

Each step needs a time limit for the user to complete it after starting the offer, and your tracking needs a window at least that long, or late conversions are not attributed. What is a conversion window? covers the details. There is also a trade-off in when you confirm: paying on a later event (a trial that converts to paid, a purchase that is not refunded) lowers your risk, but users wait longer for their reward, which can reduce completions. A common compromise is a small early step plus a larger confirmed step.

Reversals

Agree in advance which conversions you can reverse, for which reasons (fraud, chargebacks, refunds), and for how long after the conversion. Build expected reversals into your value estimate rather than reversing honest users who simply churned. What is a reversal rate? explains how the rate is measured and why publishers watch it closely.

Frequently asked questions