There is no right Meta Ads budget for subscription apps in general, but there is a right range for yours, and you can calculate it. Spend at least enough to give each ad set about 50 optimisation events a week, which is when Meta says ad sets usually leave the learning phase, and enough to read a steady flow of new creative concepts. Spend no more than the point where the next step of budget buys subscribers above your target CAC, and no more than you can fund for a quarter before store payouts catch up. For a hypothetical health and fitness app paying £15 per trial start and targeting a £45 CAC, that range is about £3,250 to £8,000 a month. Change the cost per trial or the target and the range moves a long way, which is the point.
Most budget advice starts from the wrong end. It asks what other apps spend, or what share of revenue should go to ads, and works down. For a subscription app both questions mislead, because the numbers that decide whether a pound of Meta spend is worth it are yours: what an event costs, what a subscriber is worth, and how long you can wait to be paid. This guide builds the budget from those, then shows what to do when the numbers say your app cannot yet afford the campaign it wants.
Key takeaways
- A Meta budget is a range with two floors and two ceilings, not a single number copied from another app.
- The delivery floor is roughly 50 optimisation events per ad set per week. Which event you optimise for changes it more than anything else.
- The economic ceiling is set by marginal CAC, the cost of the subscribers the last budget step bought, not by the account average.
- Subscription revenue arrives late, so cash can cap the budget before economics does.
- When a floor sits above a ceiling, change the campaign, not the maths: a shallower event, fewer ad sets, or a better funnel.
Why there is no standard Meta budget for a subscription app
We have not found a credible public benchmark for how much subscription apps spend on Meta at a given size, and we are not going to invent one. Even if one existed, it would describe what other apps chose to do, not what worked for them, and their cost per trial, prices and runway are not yours.
The common rule of thumb, spending a fixed percentage of revenue on marketing, fits a subscription app especially badly. Revenue from a subscriber arrives over months, so this month's revenue mostly reflects subscribers you acquired long ago. A percentage rule cuts spend just as new cohorts start paying back and raises it on the strength of old ones. It also says nothing about whether the next pound buys a subscriber worth having.
The right Meta budget for a subscription app is the range between the least it can spend and still learn, and the most it can spend before the next subscriber costs more than they are worth or the cash runs out.
The budget range: two floors and two ceilings
Four limits set the range. Two are floors, below which the money teaches the algorithm and you very little:
- The delivery floor. Enough of your optimisation event for each ad set to stabilise.
- The creative floor. Enough spend to judge new concepts as fast as old ones wear out.
Two are ceilings, above which spend destroys value:
- The CAC ceiling. The spend level at which the next step buys subscribers above your target CAC.
- The cash ceiling. The most media you can fund before subscription revenue reaches your bank account.
The worked example below uses one hypothetical app throughout. It sells an annual-led subscription with a free trial. Its install to trial rate is 7% and its trial to paid rate is 37%, which sit close to the Health & Fitness medians in RevenueCat's State of Subscription Apps 2026: 6.9% of downloads starting a trial within 30 days, and 37.7% of trials converting to paid. It pays £1.05 per install, so £15 per trial start and about £40.50 per paying subscriber at its starting spend, and has set a target CAC of £45. All of these are illustrative round numbers, not client data.
Floor one: enough optimisation events for Meta to learn
Meta's Business Help Centre says ad sets leave the learning phase when they can deliver stably, which usually happens after about 50 results in the week after the ad set's last significant edit. The same page is direct about budget: it recommends a budget large enough to get enough total results, and warns that a very small or inflated budget gives the delivery system an inaccurate signal. An ad set that cannot reach the threshold shows as "Learning limited".
So the delivery floor is a simple calculation:
Delivery floor per month ≈ 50 × cost per optimisation event × 4.33 weeks × number of ad sets
The event you optimise for decides almost everything. Here is the floor for one ad set in our hypothetical app:
| Optimisation event | Cost per event | 50 events a week | Monthly floor per ad set |
|---|---|---|---|
| Install | £1.05 per install | £52.50 | about £230 |
| Start Trial | £15 per trial start | £750 | about £3,250 |
| Subscribe | about £40.50 per subscription | about £2,030 | about £8,800 |
Optimising for installs is almost free to stabilise and tells the algorithm the least about who pays. Optimising for subscriptions is the most aligned with revenue and costs nearly three times as much to stabilise as trials. Each extra ad set multiplies the floor, so running the UK and the US as separate trial-optimised ad sets doubles it to about £6,500. That is why consolidation matters on a small budget, and why Meta's own guidance says that combining similar ad sets combines their learnings. We go deeper on the mechanics in our explainer on the Meta Ads learning phase, and on automated setups in our Advantage+ app campaigns guide.
Floor two: enough spend to read new creative
A budget that stabilises delivery can still be too small to learn which ads to run. New concepts need their own results before you can tell a winner from noise, and in most subscription categories winners wear out within weeks. Our guide to how much ad spend an app needs before hiring an agency sets out the formula we use: reading about four new concepts a month, with half the budget on testing, takes roughly 200 times your cost per trial. At £15 per trial, that is about £3,000 a month. The underlying per-concept numbers are in our Meta Ads creative testing budget guide.
The two floors overlap rather than add up when tests run inside the main campaign, so take the higher of the two. If you test in a separate campaign, add them. Either way, note that Meta treats adding a new ad to an ad set as a significant edit that restarts learning, so a steady stream of new creative is one more reason a budget sitting exactly on the delivery floor tends to wobble.
Ceiling one: where marginal CAC passes your target
Spend does not buy subscribers at a constant price. The first pounds reach the people most likely to convert; each extra pound reaches people a little less likely to. The question for the budget is not what the average subscriber costs but what the marginal CAC is: the extra spend in a step divided by the extra subscribers it bought.
Here is a hypothetical response curve for our app, read on matured cohorts. The target CAC is £45, set the way we describe in our guide to setting a target CAC for a subscription app.
| Monthly spend | New subscribers | Blended CAC | Marginal CAC of the step | Against £45 target |
|---|---|---|---|---|
| £4,000 | 100 | £40.00 | £40.00 (first step) | Under target |
| £8,000 | 190 | £42.11 | £44.44 | Under target |
| £12,000 | 265 | £45.28 | £53.33 | Over target |
| £16,000 | 325 | £49.23 | £66.67 | Well over target |
The ceiling is about £8,000. Look at the £12,000 row, because it is where budgets usually go wrong. The blended CAC is £45.28, close enough to target that the dashboard looks fine. The £4,000 that took the account from £8,000 to £12,000 bought 75 subscribers at £53.33 each, nearly a fifth over what the app decided a subscriber is worth. Every month at £12,000 spends £4,000 buying customers at a loss, hidden inside a healthy-looking average.
Set the budget ceiling where marginal CAC meets your target, not where blended CAC does. The average always flatters the last step.
Two cautions. First, you do not get this curve in advance; you discover it by stepping the budget and reading each step's cohort. Second, read it late enough. RevenueCat finds that 50.6% of paid conversions happen on the day of install but 19.2% arrive in week six or later, both as shares of the conversions that eventually happen. A step judged after a fortnight will look more expensive than it is. Our guide to CAC payback for subscription apps covers how long each plan type takes to earn its cost back.
Ceiling two: the cash you can fund before payouts
You pay Meta as the ads run. The stores pay you later. Apple's App Store Connect help says payments are made within 45 days of the last day of the fiscal month in which the transaction completed, and Google Play pays out around the 15th of the following month. In practice that means about two months of media leave the business before the first proceeds from it arrive, even for an annual plan that collects a full year up front. Monthly plans are slower again, because the cohort pays back over several billing cycles.
Because a fair read of any budget level takes weeks, commit to a level only if you can hold it for a quarter. At £8,000 a month that is £24,000 of media before you can judge the result. Compare two hypothetical owners of the same app. One has raised a seed round and holds £400,000; £24,000 is 6% of its cash and the CAC ceiling binds first. The other is bootstrapped with £30,000 in reserve; £24,000 would be 80% of it. If that founder decides £12,000 is the most the business can put at risk over a quarter, the cash ceiling is £4,000 a month, half the economic ceiling, and that is the budget.
Worked example: putting the range together
Here are all four limits for the hypothetical app, plus the one campaign it cannot yet afford:
| Limit | Monthly spend | Why |
|---|---|---|
| Floor: Meta delivery (trial-optimised, one ad set, one market) | about £3,250 | 50 trial starts a week at £15 each. |
| Floor: reading new creative | about £3,000 | Four new concepts a month, using the formula in our hiring-spend guide. |
| Ceiling: marginal CAC | about £8,000 | The last step that bought subscribers under the £45 target. |
| Ceiling: cash | depends on the business | A funded app clears it easily. A bootstrapped one may not. |
| Not affordable yet | about £8,800 | The floor for a purchase-optimised ad set sits above the CAC ceiling. |
The funded version of this app should spend between about £3,250 and £8,000 a month on one market, optimising for trial starts. A sensible plan is to start near £5,000, which clears both floors with some headroom, then step up towards £8,000, reading each step on matured cohorts before taking the next. The bootstrapped version should run £4,000 on one market and add no second market until the first payouts arrive.
Notice what the range rules out. Even at the CAC ceiling, the app buys about 190 subscribers a month, or about 44 a week, which is below the 50 a week a purchase-optimised ad set needs. Optimising for subscriptions would mean spending above the ceiling just to feed the algorithm. Until the funnel or the budget changes, trial starts are the right event.
What to do when the floor is above the ceiling
Sometimes the least you can spend and still learn is more than you can afford. Spending anyway in the hope it works out is the expensive option. These are the better ones, roughly in order:
- Use a shallower event. Move from Subscribe to Start Trial, as in the example. The floor drops by more than half while the event still sits close to revenue.
- Run fewer ad sets. One broad ad set in one market has a quarter of the floor of four split ones.
- Choose a cheaper market to learn in. Only if its users behave like the market you actually want, and you check that rather than assume it.
- Fix the funnel first. A better install to trial or trial to paid rate lowers every floor and raises the CAC ceiling at the same time.
- Treat it as research. Run a capped learning test judged on what it teaches you, as we describe in our paid acquisition readiness checklist, rather than a campaign judged on return.
Working backwards from a growth target
Founders often arrive with a target instead: we want 400 new subscribers a month. That is a useful check on the range, not a way to set it. Multiplying 400 by the £40 the first subscribers cost suggests £16,000. The response curve says £16,000 actually buys about 325 subscribers, with the last step costing £66.67 each. Hitting 400 inside the target is not possible on this funnel and this creative, however the budget is set.
When the goal and the ceiling disagree, the levers are the ones that move the curve rather than the spend. Lifting trial to paid from 37% to 42% cuts the cost of every subscriber on the curve by about 12%, which brings the £12,000 step down from £53.33 to about £47, almost inside target. Stronger creative shifts the whole curve down. A new market adds a second curve that starts cheap again. Our break-even CPI guide shows how each conversion rate feeds through to what you can pay for an install.
How to raise a Meta budget without resetting learning
Meta's page on significant edits and the learning phase says a budget change may or may not be significant depending on its size. Its own example: moving from USD 100 to USD 101 is unlikely to send ad sets back into learning, while moving from USD 100 to USD 1,000 may. The learning phase page separately advises against frequent budget changes. Meta does not publish a safe percentage, so treat any rule you read elsewhere as someone's habit, not a platform rule.
What follows from that is a step plan. Decide the steps in advance, make each one large enough to show up in the cohort data, hold it long enough to read, and take the next only if the last one came in under target. In the example that means steps of £1,500 to £3,000 from a £5,000 start. Going straight from £5,000 to £12,000 would likely restart learning and also skip the reading that would have shown the £12,000 level is unprofitable.
What moves the ceiling over time
The range is not fixed. Three things move the CAC ceiling up or down from month to month:
- Creative. When winners tire, the same spend buys subscribers at a higher price and the ceiling falls. Our guide to creative fatigue in mobile app ads covers the signs, and the creative refresh calculator estimates how much new creative a spend level needs.
- The funnel. A new onboarding flow, paywall or price can move every CAC on the curve, in either direction.
- Measurement. On iOS, attributed conversions arrive late and incomplete, so read the curve from your own subscription data. Our post on SKAdNetwork postback delays explains why Meta's numbers and yours will differ.
Our Steps & Beasts case study is an example of the funnel and the creative moving together. The published result is a 1112% revenue increase and 118% more active subscriptions, and the case study credits heavy creative testing alongside onboarding redesigns. Changes like that are what let a budget grow without breaching its ceiling.
Where we fit
Our pricing page sets a £3,000 minimum monthly ad spend, because below that an account cannot read a creative test reliably. That lines up with the floors above at a typical cost per trial. If your range sits above that and the constraint is making enough strong creative or finding the CAC ceiling without overshooting it, our Meta Ads service runs creative strategy, production and campaign management as one team. You can see how we work with consumer and subscription apps, and if your app is already spending, you can apply to work with us. For how the budget fits into the rest of the funnel, start with our subscription app marketing strategy guide.
Frequently asked questions
How much should a subscription app spend on Meta Ads per month?
Enough to clear two floors and no more than two ceilings allow. The floors are the spend that gives each ad set about 50 optimisation events a week, which is when Meta says ad sets usually leave the learning phase, and the spend needed to read new creative concepts. The ceilings are the point where the next step of spend buys subscribers above your target CAC, and the amount of media you can fund before store payouts arrive. In our hypothetical example the range came out at roughly £3,250 to £8,000 a month, but your own cost per trial and target CAC will move it a long way.
What is the minimum Meta Ads budget for a subscription app?
The practical minimum is the spend that buys about 50 of your optimisation event per ad set each week. Multiply 50 by your cost per event and by about 4.33 weeks. At a hypothetical £15 per trial start that is about £3,250 a month for one ad set optimising for trials. Optimising for paid subscriptions needs 50 subscriptions a week instead, which in this example costs nearly three times as much.
Should a subscription app spend a percentage of revenue on Meta Ads?
A percentage of revenue is a poor way to set the budget for a subscription app. Revenue arrives months after the spend that produced it, so the rule cuts spend while new cohorts are still paying back and raises it based on subscribers acquired long ago. Set the budget from what the next pound of spend buys, measured as marginal CAC on matured cohorts, and from the cash you can fund before payouts arrive.
How quickly can I increase my Meta Ads budget?
Meta says a budget change may or may not reset the learning phase depending on its size, giving the example that USD 100 to USD 101 is unlikely to, while USD 100 to USD 1,000 may. It also advises avoiding frequent budget changes. Raise spend in steps you can measure, hold each step long enough to read the cohort it bought, and only take the next step if the last one came in under your target CAC.
Is it better to optimise for trials or purchases on a small Meta budget?
On a small budget, usually trials. A purchase-optimised ad set needs about 50 subscriptions a week to stabilise, and if the budget cannot buy that, the ad set spends most of its life in the learning phase. Trial starts are cheaper and more frequent, so the same budget can feed the algorithm. Move deeper once the budget can buy enough purchases without going above your target CAC.
The short version
Work the budget out from your own numbers, not from what other apps spend. Find the least that gives each ad set about 50 optimisation events a week and lets you read new concepts. Find the most before the next step of spend buys subscribers above your target CAC, or before the cash runs short of the payouts. Spend inside that range, on the deepest event the range can afford, and raise the budget in steps you read on matured cohorts. If the floor sits above the ceiling, change the campaign or the funnel rather than spending through it.