Weekly vs annual subscriptions: what your plan mix does to ad payback

Two plans, two payback curves

A payback curve shows how much of your ad spend has come back, day by day, from the customers the ads brought in.

An annual plan collects a year in one payment, so the curve jumps when each customer pays and then stays flat until the renewal. A weekly plan collects a week at a time, so the curve starts low and climbs with each renewal, slowing as subscribers cancel. Plan mix, the split between plans, sets the shape.

A worked example: one budget, two plan mixes

Take a fictional app on the App Store, outside Apple's Small Business Program, with a weekly plan at $7.99 and an annual plan at $49.99 before tax, both with a 3-day free trial. After the 30% first-year cut, a weekly payment brings in $5.59 and an annual one $34.99. Of weekly subscribers, 55% renew after their first paid week, then 92% of those still paying renew each week.

Two campaigns each spend $10,000 in the same week and win 300 new paying customers. In campaign W, 240 pick weekly and 60 annual. In campaign A, it is the reverse. The table shows their proceeds after the store's cut, and as a multiple of spend. Day 7 means day 0 to day 6.

At day 7, A looks two and a half times better, and it pays back on day 73 against day 108 for W. By day 365, W has brought in 14% more, because its weekly subscribers kept paying while A's annual subscribers paid once.

What day 7 and day 30 can tell you

Early ROAS shows how much of its year a campaign has collected. Hold the campaigns above to a day-7 target of 0.5× and you scale A and cut W, the one that earns more over the year. Day 30 does not fix it: 0.55× against 0.92×. Our break-even guide shows why one day-7 target misleads across apps. Plan mix does the same within one app.

Ads Manager tilts further. Meta's default attribution setting for new ad sets is 7-day click or 1-day view (Meta). RevenueCat's Meta integration sends renewals as Subscribe events (RevenueCat), but a weekly renewal comes after a 7-day window has closed, so Ads Manager can credit an annual plan's full price and only a weekly plan's first payment.

Annual renewals take a year to judge

Annual subscribers pay for their first year as soon as their trials convert, but renew only a year later, so a campaign you start this week shows its renewal rate next year.

That rate can change the verdict. Each renewal in the example brings in $42.49, since Apple's cut drops to 15% after a year of paid service. If 30% of A's 240 annual subscribers renew, A gains $3,059 and pulls ahead, even counting W's own renewals. At 10%, it gains $1,020 and W stays ahead. With the same rate on both, the line sits at about 20%.

Early hints exist. Apple's server notifications tell you when a subscriber turns auto-renew off or back on (Apple), and RevenueCat's Cohort Explorer counts each cohort's subscriptions still set to renew (RevenueCat). That share is your earliest read on renewals, though people can change their minds. Until your annual cohorts have renewed once, treat any second-year forecast as a wide range.

Refunds and early cancellations

Each refund takes back one payment, or part of it, so refunds weigh more on annual plans. On the App Store, customers ask Apple for refunds and Apple decides (Apple). On Google Play, buyers can ask Google within 48 hours of a purchase and are pointed to the developer after that (Google), and you can issue full or partial refunds from Play Console (Google).

If 5% of first payments in the example were refunded, A would lose $437 in proceeds and W $172. RevenueCat's Cohort Explorer deducts a refund when it happens, not on the purchase date (RevenueCat), so a strong day-7 figure can shrink by day 30.

Early cancellations work differently. On Google Play, a subscriber who cancels keeps access for the period already paid (Google), and App Store subscriptions renew at the end of each period until the user cancels (Apple). A weekly subscriber who cancels in week two never pays for week three. An annual one has already paid for the year, so the loss stays hidden in revenue until the renewal date.

The store's cut after the first year

On the App Store, outside the Small Business Program, you receive 70% of each subscription payment in a subscriber's first year and 85% once they have a year of paid service. Free trials do not count toward that year, and Small Business Program members receive 85% from the first payment (Apple).

In year one both plans pay the same rate, so the cut scales both curves alike: in the Small Business Program, every figure in the table would be about 21% higher. After a year, every annual renewal gets the lower rate, but on a weekly plan only the subscribers still paying do, fewer than 1 in 100 in the example.

Google Play has no such step. It takes 15% of auto-renewing subscriptions, or in the US, EEA, UK, Australia and Japan a 10% service fee plus a 5% billing fee (Google). Other App Store rates, such as the EU's, are in our break-even guide.

Cash flow and how fast you can reinvest

Proceeds are not yet cash. Apple pays within 45 days of the end of the fiscal month of the sale (Apple), and Google Play pays each month's orders around the 15th of the next month (Google). Ad spend is billed as you go.

Plan mix decides how much of that gap you fund. Leaving the payout delay aside, by day 30 W has earned back $5,484 of its $10,000 and A $9,244, so W leaves $4,516 to cover from other cash and A only $756.

An annual-heavy mix lets you put the same dollars back into ads sooner, which matters most when cash limits your budget. But sooner is not more: W still earns more by day 365, and refunds can still take back part of A's cash.

How to compare campaigns with different plan mixes

Compare campaigns on what their customers will pay over the year. Our break-even guide turns early revenue into a forecast with one multiplier from past cohorts, which works while the plan mix holds steady. When campaigns differ in mix, use one value per plan:

In the example, weekly payers bring in $43.50 each in year one and annual payers $34.99, so W's forecast is 240 × $43.50 + 60 × $34.99, or $12,539. On proceeds, break-even sits just above 1.0×.

1. Split new payers by plan. Count payers, not trials, by the plan they first bought. RevenueCat's charts can be segmented by product duration (RevenueCat), and App Store Connect's analytics can filter subscriptions by plan and duration (Apple).
2. Value each plan from mature cohorts. For cohorts a year old or more, take each plan's proceeds per payer by day 365, after refunds, and keep the lowest and highest values for a range. RevenueCat's Cohort Explorer shows proceeds by cohort (RevenueCat).
3. Compare like with like: the same store, countries, weeks, prices and paywall.
4. Check early signals. If first-month refunds, the first weekly renewal or the share of annual subscribers still set to renew stray from past cohorts, widen the range.

Without an MMP, you rarely know each customer's campaign. A campaign running alone in a country or on one platform shows its mix there. Otherwise, count new payers by plan above your baseline on the days it spends.

What to test on your pricing page

Your paywall shapes the plan mix too. What works differs by app, audience, country and season, so treat these as questions to test, not answers:

• Which plan is preselected or highlighted.
• How the annual price is shown: alone, or with a weekly equivalent beside it.
• Where the trial sits: on the annual plan, the weekly plan, both or neither.
• How many plans you show, and whether a monthly plan takes buyers from weekly or from annual.
• The gap between the weekly and annual prices.

Randomize who sees which version. RevenueCat's Experiments A/B test prices, trials and the number and mix of products, and follow each variant through the subscription lifecycle (RevenueCat). Judge variants on forecast year-one proceeds per person who saw the paywall, after refunds and the store's cut. A paywall that moves buyers from weekly to annual can look like a win at day 7 whether or not it is one.

Mind the store rules. Apple requires each subscription's length and price to be disclosed clearly and conspicuously (Apple), and Google Play's policy lists annual plans that most prominently show a monthly cost as a violation (Google). Keep the billed amount easy to see.

How INTARC does this

INTARC reads your revenue from App Store Connect, Google Play or RevenueCat, and your ad accounts, with read-only keys and no SDK. It follows each week's new customers by what they actually paid, plan by plan, after the store's cut, and forecasts whether each week of spend will pay back within a year, with a range. Every forecast it makes on its own studio's apps is written down in advance and scored in public at /proof. You can look around the demo or create an account to run it on your own apps.

Frequently asked questions

Is a weekly or an annual plan better for paid ads?

Neither in general. Annual plans return cash sooner and bring a renewal a year later, while weekly plans can bring in more over the first year if subscribers keep renewing. Measure both in your own cohorts.

Why does my annual-heavy campaign look so much better at day 7?

Its annual subscribers have paid for their first year, while weekly subscribers have paid for one week. Day-7 revenue shows what a campaign has collected so far, not what it will collect.

When can I judge an annual campaign's renewals?

Only when they happen, a year after each subscriber's first payment. Until then, refunds and the share still set to renew are your early signs.

Does Apple's 15% rate after one year apply to weekly plans?

Yes, once a subscriber has a year of paid service in the same subscription group, whatever the plan. Free trials do not count toward the year, and Small Business Program members pay 15% from the first payment.

How do I compare campaigns with different plan mixes?

Count each campaign's new payers by plan, value each plan by what its payers brought in over a year in your older cohorts, and compare the total with spend.

Store fees, refund rules and reporting tools change often. We checked every linked source on November 13, 2026. Examples use fictional apps and numbers.