3-day, 7-day or no free trial: how trial length changes ad measurement
A 3-day, 7-day or no free trial moves the first payment in or out of Meta's and SKAdNetwork's windows. How to read campaigns and test trials fairly.
Key takeaways
- Trial length decides when the first payment lands: at once with no trial, and when the trial ends with a 3-day or 7-day trial.
- Against Meta's default of 7-day click or 1-day view, a 3-day trial's payment usually counts and a 7-day trial's often does not, and SKAdNetwork campaigns show no trial payments in Ads Manager.
- Optimize for the deepest event each ad set gets about 50 times a week, but judge every campaign on payers.
- Read trial starts as the early signal, then count payers from the same start days a trial length later.
- Compare a 3-day and a 7-day trial on randomly split new users in the same weeks and countries, by payers per new customer after refunds.
Trial length moves the first payment
A free trial does not change when your campaigns spend. It changes when the money arrives. With no trial, the first payment comes at the purchase. With a trial, it comes when the trial ends: App Store subscribers are not billed until the free trial is over, and the subscription then renews at the regular price (Apple).
Ad networks credit only what happens inside fixed windows, so trial length decides what they see. For the general reasons the dashboards disagree, see why Meta Ads Manager and RevenueCat never match.
| Setup | First payment | Meta, 7-day click | SKAdNetwork 4 |
|---|---|---|---|
| No trial | At purchase, often day 0 | Inside | First window |
| 3-day trial | About day 3 | Inside | Second window |
| 7-day trial | About day 7 | At or past the edge | Second or third window |
Meta's attribution window and the first payment
Meta's default for new ad sets is 7-day click or 1-day view, and Ads Manager files each conversion under the date of the impression (Meta).
- No trial. The purchase usually comes soon after the click, so it counts.
- 3-day trial. The payment comes about three days after the trial starts, inside the click window.
- 7-day trial. The payment comes about seven days after the trial starts, at or just past the window's edge, so Ads Manager often shows the trial and misses the payment.
With either trial, the 1-day view window misses the payment unless the person saw one of your ads in the day before paying. On iOS, app campaigns that use Meta's own attribution count events after the install on click only, not on view (Meta). And Meta counts only payments it receives: the Meta SDK cannot reliably report subscription events that happen while the app is closed, while a server integration can (RevenueCat).
So treat Ads Manager's most recent week as unfinished, since payments keep landing under past impressions, and expect its purchases to fall when you move from a 3-day to a 7-day trial, even if real payers rise.
SKAdNetwork's conversion windows
SKAdNetwork 4 has three conversion windows, counted from the first launch: days 0 to 2, 3 to 7 and 8 to 35. Only the first postback can carry a fine-grained conversion value, and the later two carry a coarse one (Apple). Ads Manager shows only the first postback with a fine value, while the later two appear only in Meta's Ads Insights API (Meta).
So with no trial, a purchase in days 0 to 2 can reach Ads Manager. With a 3-day or 7-day trial, only the trial start fits in the first window, and the payment never shows there. Server events cannot fix this, because the app itself updates the conversion value (Apple). Let the first window record the trial start, and read payments from RevenueCat or the stores. Our SKAdNetwork 4 guide shows how to design the values.
Optimizing for trial starts or purchases
An app promotion campaign can optimize for an app event you choose, such as a purchase (Meta). With a trial, the usual choice is the trial start or the payment.
- Trial starts usually come minutes after the install, inside every window and the first SKAdNetwork postback, and in volume. The risk: Meta learns to find people who start trials, not people who pay.
- Purchases are closer to money but fewer and later. With a 7-day trial many fall outside the window, so Meta learns from a thin, late signal.
- No trial makes the purchase the first event, but fewer people usually buy outright than start a free trial.
Meta says an ad set usually leaves the learning phase after about 50 results in the week after its last significant edit (Meta), and suggests a more frequent optimization event when it cannot get there (Meta). So optimize for the deepest event each ad set reaches about 50 times a week, and judge it on payers either way.
Read trial starts early, payers a trial length later
Trial starts are your earliest signal, and payers are the verdict.
- Day 1 or 2: cost per trial start. Compare each campaign with its own recent weeks. A jump is worth a look, not a verdict.
- Forecast payers from trial starts and the trial-to-paid rate of recent complete weeks.
- Two or three days after the trial ends: count payers whose trials started on the days you are judging, not everyone who paid this week.
- Value them. Compare cost per payer with what a payer brings in after the store's cut, as in our break-even ROAS guide.
cost per trial = spend ÷ trial starts
trial-to-paid = payers ÷ trial starts, same start days
cost per payer = cost per trial ÷ trial-to-paid
expected payers = trial starts × recent trial-to-paidCount again a week or two later. If a first charge fails, the App Store keeps trying for up to 60 days (Apple), so some payers arrive late.
Trial-to-paid rate by campaign
Cost per trial alone misleads. A fictional campaign paying $8 per trial, with 40% converting, pays $20 per payer. One paying $6 per trial with 25% converting pays $24.
- From Ads Manager, divide subscriptions by trials started. This works with a 3-day trial. With a 7-day trial the window clips payments, and SKAdNetwork campaigns show none.
- From your revenue data, filter by country. RevenueCat's charts filter by country, store, product and offer, but not by Meta campaign (RevenueCat), so give campaigns their own countries where you can.
Judge each rate against your own complete cohorts, not a benchmark: a good rate depends on your price, trial, countries and paywall.
A worked example: the same spend, two trial lengths
Picture one fictional week of Meta spend twice, once with each trial: $1,000 a day for seven days, and 400 installs a day either way. Every trial starts on install day and pays exactly a trial length later. Day 7 is the last day of spend, and refunds are left out.
| 3-day trial | 7-day trial | |
|---|---|---|
| Ad spend, days 1 to 7 | $7,000 | $7,000 |
| Installs | 2,800 | 2,800 |
| Trial starts | 700 | 630 |
| Cost per trial start | $10.00 | $11.11 |
| Payers by day 7 | 136 | 0 |
| Payers by day 10 | 238 | 108 |
| Payers by day 14 | 238 | 252 |
| Trial-to-paid | 34% | 40% |
| Payers per 1,000 installs | 85 | 90 |
| Cost per payer | $29.41 | $27.78 |
On day 7, the 3-day trial shows 136 payers at about $51 each, and the 7-day trial shows none. By day 14, the 7-day trial has 14 more payers and costs $1.63 less per payer. Its trial-to-paid rate looks far better, 40% against 34%, but per install the gap is 90 payers per 1,000 against 85, because fewer people started the longer trial.
Ads Manager would favor the 3-day trial even longer. The numbers only show the timing: your own test may favor either trial.
How to compare a 3-day and a 7-day trial fairly
Comparing this month's trial with last month's mixes the trial up with the season, creatives, prices and country mix. Run both at once.
- Split new users at random between the two trials, in the same weeks. RevenueCat's Experiments, on its Pro and Enterprise plans, can A/B test trial length and trial presence (RevenueCat).
- Keep countries, platforms and campaigns the same in both arms, and read large countries on their own too.
- Count by cohort. Group customers by the day they first opened the app, not the day they paid. RevenueCat's Conversion to Paying chart does this, can limit each cohort to its first 3, 7, 14 or 30 days, and drops first payments refunded within that limit (RevenueCat). Use 14 days or more, so the 7-day arm is not cut off.
- Compare payers per new customer, then revenue after refunds. Trial-to-paid alone misleads, because each arm divides by its own trial starts.
- Wait for the first renewal before you decide.
How Apple and Google Play set up free trials
On the App Store, a free trial is one of three types of introductory offer, with pay as you go and pay up front. It can last from 3 days to a year, in fixed steps of days, weeks or months. Each person gets one introductory offer per subscription group, and each subscription has one current offer per storefront (Apple), so testing two trial lengths at once takes two products.
On Google Play, each base plan of a subscription can carry offers. An offer can start with a free trial of 3 days to 3 years, and when someone qualifies for several offers, your app decides which to show (Google). One base plan can carry both a 3-day and a 7-day offer.
Cancellations and refunds
People who decide not to pay usually turn off renewal during the trial. Apple tells App Store customers to cancel at least 24 hours before a trial ends if they do not want it to renew (Apple). RevenueCat's Trial Conversion chart splits running trials into set to convert and set to cancel (RevenueCat): an early, partial read on each cohort.
Refunds come after the first payment. App Store customers ask Apple, which decides (Apple), while on Google Play you can refund orders in Play Console (Google). RevenueCat's Trial Conversion chart still counts a refunded first payment as a conversion (RevenueCat), so subtract refunds yourself. In a trial test, compare the arms' refund rates too: RevenueCat's Refund Rate chart shows what share of a period's payments were later refunded (RevenueCat). Extra payers who take their money back are not extra payers.
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 counts the paying customers each campaign adds on the days and in the countries it spends, so a trial that converts after an ad network's window still counts. It forecasts whether the 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 3-day or a 7-day free trial better for paid ads?
Neither in general. A 3-day trial shows its payments sooner and inside Meta's default window, so it looks better early. A 7-day trial can end with as many payers or more. Only a split test tells you which wins for your app.
Why did Ads Manager purchases drop after we moved to a 7-day trial?
Probably because the first payment moved from about day 3 to about day 7, at or past the edge of the 7-day click window. Check payers in RevenueCat or the stores, by cohort, before you decide the trial hurt sales.
Should Meta campaigns optimize for trial starts or purchases?
Optimize for the deepest event each ad set gets about 50 times a week, which with a 7-day trial is often the trial start. Either way, judge campaigns on cost per payer, a trial length later.
When can I judge a campaign that uses a free trial?
Read cost per trial start after a day or two. Count payers two or three days after the trial ends, and again a week or two later for late charges and refunds.
Is it easier to measure ads without a free trial?
Yes. The purchase happens at once, inside every attribution window and the first SKAdNetwork postback, so Ads Manager can see it and Meta can optimize for it. Whether no trial earns more is a question for a test.
Store rules, attribution settings and chart definitions change often. We checked every linked source on November 13, 2026. Examples use fictional apps and numbers.