Run Meta traffic straight to the App Store or Google Play unless you have a specific reason to put the purchase on the web. A web-to-app funnel, where the ad leads to a web page that takes payment before the user downloads the app, saves you the store commission. It also usually loses buyers at the payment step. The trade is simple to state: against Apple's 30% first-year commission and about 6% in web payment costs, a web checkout can convert roughly 25% fewer buyers and still break even. Against a 15% commission, it can afford to lose only about 10%. Apps with long, persuasive onboarding, higher-priced annual plans and mostly US revenue are the strongest candidates to clear that bar. Apps whose value shows in the first minute of use usually do not.
This guide treats the choice the way it shows up in an ad account: as a question about net revenue per pound of spend, what Meta can measure on each route, and what the stores charge on each path as of November 2026. It is not a fees explainer. The fees matter, but they are the smaller half of the decision.
Key takeaways
- "Web-to-app" covers four different funnels. Only a web checkout changes what the stores charge.
- A web checkout wins only if it keeps enough buyers. Divide (1 minus store fee) by (1 minus web costs) to find how many.
- Passing the fee saving to customers as a web discount spends the margin before you have earned it back in volume.
- The web gives Meta a purchase signal in the same session. It does not make later events like trial-to-paid arrive any faster.
- Never compare Meta-reported ROAS across the two routes. Compare net revenue per pound on cohorts of the same age.
The four funnels people mean by "web-to-app"
Much of the confusion in this debate comes from one label covering several different journeys. Before comparing anything, decide which of these you are actually considering.
| Route | Journey | Meta setup | Store fee on the purchase | Main risk |
|---|---|---|---|---|
| 1. Direct to store | Ad → App Store or Google Play → install → onboarding and paywall in the app | App campaign optimising for installs or an app event (Start Trial, Subscribe, Purchase) | Standard store commission on every payment | iOS measurement is aggregated and late |
| 2. Web pre-lander | Ad → your web page or quiz → store → install → paywall in the app | Website campaign optimising for a web event such as a quiz completion or lead | Standard store commission (the purchase still happens in the app) | An extra step that leaks buyers without saving any fees |
| 3. Web checkout (web-to-app) | Ad → web onboarding → web paywall and payment → download the app to use it | Website campaign optimising for a Pixel event such as Subscribe or StartTrial | No store commission on the web payment | Fewer buyers at the payment step; you become the merchant |
| 4. App-to-web link-out | Ad → store → install → link inside the app → web checkout | App campaign as in route 1; the purchase leaves the app afterwards | Depends on store and region (see the fee table) | Rules and rates are still moving, especially in the US |
Route 2 is the one to be most sceptical of. A web pre-lander that sends people on to the store adds a step without removing the commission, so it has to earn its place some other way: by capturing emails, by testing messages faster than the store listing allows, or by feeding Meta an earlier event. Sometimes it does. Often it is a web funnel in name only.
Route 4 is not really a Meta decision at all. The ad still drives an install, and the choice to send buyers from inside the app to the web is a product and legal decision made after the click. It is covered here because its rules changed this year and it is the route most often confused with a true web funnel. The rest of this guide is mostly about routes 1 and 3.
What Apple and Google charge on each path in November 2026
Store fees depend on where the purchase happens, not where the ad sent the user. This table is current as of 4 November 2026. The US link-out rules are in active litigation, so check them again before you build anything that depends on them.
| Where the purchase happens | United States | European Union | UK |
|---|---|---|---|
| In-app purchase (routes 1 and 2) | Apple 30% in year one, 15% after a year of paid service or in the Small Business Program. Google Play 15% on subscriptions. | Apple 26%, or 15% for most small developers and for subscriptions after year one (from 1 Oct 2026). Google Play 15% on subscriptions. | As the US for Apple. Google Play 15% on subscriptions. |
| Web purchase started from an ad (route 3) | No store commission. The app must also offer the subscription as an in-app purchase unless it qualifies for a separate exception. | No store commission. Same in-app availability rule. | No store commission. Same in-app availability rule. |
| Link from inside the app to web checkout (route 4) | Apple: currently barred from charging commission by court order; rate under review, Apple has proposed 15%. Google: 10% on recurring subscriptions within 24 hours of the link, payable from 1 Oct 2026. | Apple: 15%, or 10% for developers in its reduced-rate programmes (from 1 Oct 2026). Google: separate EEA programmes. | Apple: not generally permitted; only in storefronts where Apple offers a link-out entitlement. Google: check the Play Console programmes for the UK. |
The web checkout row rests on Apple's App Review Guidelines. Rule 3.1.3(b), on multiplatform services, allows an app to give users access to subscriptions they bought on your website, provided those items are also available as in-app purchases within the app. So a web checkout does not let you remove in-app purchase. It lets you sell the same subscription somewhere Apple does not take a share.
The US link-out position comes from Apple's 2026 annual report (Form 10-K). A court injunction from April 2025 stops Apple charging any commission on purchases made through external links in US apps. In December 2025 the Ninth Circuit upheld it in part, but ruled that Apple may charge some commission on link-out purchases and may require links to match its in-app purchase option in size, form and placement. The rate went back to the district court, and the Supreme Court agreed in June 2026 to review the contempt standard. In August 2026 Apple proposed a link-out commission of 15% for standard apps, 10% for subscription renewals and 5% for Small Business Program members. Epic called the fees "far outside of the bounds" of the court's guidance. Treat today's 0% as temporary.
Google's US external content links programme charges a service fee on transactions completed within 24 hours of a user following an in-app link: 10% on recurring subscriptions, with reporting and payment required from 1 October 2026. In the EU, Apple's new terms from 1 October 2026 set a 15% commission on purchases made through links out of the app, or 10% for developers in its reduced-rate programmes. Both of those fees attach to links inside the app. A purchase that begins with a Meta ad and ends on your website never passes through one.
The break-even rule: how many buyers can a web checkout lose?
The fee saving is easy to calculate and easy to overrate, because it is only half the equation. The other half is how many people complete a web payment compared with a native store purchase sheet, which is one or two taps with a saved card and Face ID.
For the same ad spend and the same price, a web checkout matches the store route on net revenue when:
web buyers ÷ store buyers = (1 − store commission) ÷ (1 − web payment costs)
Using about 6% for web costs, which is the blended figure RevenueCat measured for its own app once processing, chargebacks and other costs were included, the ratios work out like this. Against a 30% commission, 0.70 ÷ 0.94 is 0.745, so the web can convert about 25% fewer buyers and still match the store. Against 15%, 0.85 ÷ 0.94 is 0.904, so the web can lose only about 10%. That second case covers apps in Apple's Small Business Program and every Apple subscription after its first year of paid service, which is why the web looks much less attractive to a small app or one with strong renewals.
A worked example
Here is a hypothetical app selling a £49.99 annual plan, with £49.99 treated as the price after sales tax on both routes so the comparison is like for like. A fixed Meta budget produces 200 paying subscribers through the store route. The question is what the same budget is worth on the web at different conversion rates. These are illustrative numbers, not client data.
| Route | Paying subscribers | Gross revenue | Fees and costs | Net revenue |
|---|---|---|---|---|
| Direct to store, 30% commission | 200 | £9,998 | 30% | £6,999 |
| Direct to store, 15% commission | 200 | £9,998 | 15% | £8,498 |
| Web checkout, same buyers | 200 | £9,998 | 6% | £9,398 |
| Web checkout, 15% fewer buyers | 170 | £8,498 | 6% | £7,988 |
| Web checkout, 25% fewer buyers | 150 | £7,499 | 6% | £7,049 |
| Web checkout, same buyers, 30% web discount | 200 | £6,999 | 6% | £6,579 |
At a 30% commission, the web checkout wins in every row except the last. At 15%, it loses as soon as it converts 15% fewer buyers. The last row is the trap. Many web funnels advertise a discount to "pass on the saving", and the mixed variant of RevenueCat's test offered a 30% web discount on the annual plan. With the same number of buyers, that discount leaves less net revenue than the store route even at a 30% commission. To come out ahead, the discounted web checkout needs about 6% more buyers than the store at a 30% commission and about 29% more at 15%. A discount can be a good test. It is not free money.
The same logic sits behind our guide to break-even CPI: every pound you can spend to acquire a user is set by what that user is worth after fees. Change the fee and you change the ceiling, but only if conversion holds.
What the evidence says about conversion
The best public test is RevenueCat's 2025 experiment on its own app, Dipsea, which split about 12,500 new US iOS users between in-app purchase only, a choice of in-app or web, and web only. Of the users who tapped the paywall button, about 71% went on to start a trial or subscription through in-app purchase, against 44% on the web. Conversion to paying was 6.3% for in-app only and 5.3% for web only, so the web kept about 84% of the buyers. Revenue per paying customer was also lower on the web ($39.20 against $47.09). After fees, web-only earned $1.96 per user, against $2.09 for in-app only at a 30% commission and $2.53 at a 15% commission.
Read it carefully. It was one app, in one market, measured over about three weeks, and it tested route 4, a paywall inside the app sending people to the web, not a Meta ad leading to a web checkout. A user who has already installed and opened an app has a native purchase sheet one tap away, so the web is at its biggest disadvantage there. A web funnel that sells the problem on the web and takes payment there, before the app ever enters the picture, is a different journey with a different audience. What the test does establish is that the payment step itself costs buyers, and that the fee saving alone did not make up for it.
Market-level data points the same way. RevenueCat's State of Subscription Apps 2026 puts web revenue at 3.2% of subscription app revenue globally, 4.9% in North America and 0.8% in India and South East Asia. Adoption is concentrated at the top: 41% of apps in its highest revenue tier generate web revenue, against 1.3% in the lowest. That is a correlation, not proof that the web caused their scale. Larger apps also have the engineering and finance teams that a web checkout needs.
How Meta measures each route
Measurement is the strongest argument for a web checkout, and it is often overstated. On route 1, Meta learns about installs and in-app events through the Facebook SDK or your mobile measurement partner. On iOS, Advantage+ app campaigns ask you to choose between Aggregated Event Measurement and SKAdNetwork, and Apple's SKAdNetwork reports conversions in windows of days 0 to 2, 3 to 7 and 8 to 35 after first launch, each with a random delay before the postback arrives. We cover what that lag does to decisions in our guide to SKAdNetwork postback delays.
On route 3, the purchase happens on a website you control, so it is reported through the Meta Pixel and the Conversions API, which Meta says sends events from your server that are "processed like events sent using the Meta Pixel". Meta's standard web events include StartTrial and Subscribe, so a web funnel can optimise directly for the event that matters and report it in the same session. That is a real advantage at the moment of purchase.
It has three limits. First, a web free trial still converts to paid days later, and Meta's longest standard click window is seven days, so the event that decides your revenue can still fall outside it. Second, the web campaign only sees the web. Meta has no direct view of whether the buyer downloaded the app, logged in and stayed, which is what renewal depends on, so you need your own link between the web account and the app account. Third, the two routes are now measured by different systems with different coverage and timing. Their Meta-reported ROAS figures are not comparable, and a team that moves budget towards whichever looks better in Ads Manager is usually rewarding the better-measured route, not the better one. Our guide to choosing an optimisation event sets out the volume, predictiveness and latency tests that apply on either route.
What changes when you become the merchant
The store commission pays for more than a payment form. On the web, the work moves to you. RevenueCat's web-to-app guide puts raw processing at around 2% to 3%, rising to about 6% once every cost is blended in. The costs that turn 3% into 6% are the ones teams forget to budget for: sales tax and VAT in every country you sell to, refunds, chargebacks and failed payments. A merchant of record service will take on tax and compliance for a higher fee, which narrows the saving further. Get a quote and put it in the formula rather than using the headline processing rate.
Three operational differences matter to paid acquisition directly:
- Free trials behave differently. RevenueCat notes that web buyers often enter cards with no balance, so many web funnels replace the free trial with a low-priced paid introductory period or a money-back guarantee. That changes your optimisation event and makes it harder to compare the web with an in-app free trial.
- Cash arrives sooner. Web processors usually pay out within days, while store payouts can take weeks. For an app funding growth from revenue, that can matter as much as the fee, which is why CAC payback should be measured in cash as well as in revenue.
- The store listing gets less help. A web buyer downloads the app after paying, if at all, and paid spend no longer pushes people through the store page first. Apps that rely on rankings and ratings for a large share of organic installs should treat that as a real cost.
The creative is not interchangeable
A web funnel attracts a different audience from an app install campaign, so it needs its own creative testing. In RevenueCat's 2026 report, Rise Science described web visitors as further back in the buying decision than people who install an app. Its web funnel spends longer helping people recognise the problem before it introduces the product, and it regularly finds ads that fail on app campaigns but win on the web, and the reverse. Treat that as one company's experience rather than a rule, but it matches what the mechanics suggest. An install ad has to earn a tap to a store page. A web funnel ad has to earn the first step of a longer argument that ends in a payment form.
The practical consequence is that you should not judge a web funnel on ads that were built to win installs. If the first web screen does not continue the exact promise the ad made, you have tested a message mismatch, not a funnel. Our guide to hard paywall vs freemium makes the related point for in-app paywalls: the more of the product's value an ad can carry, the earlier you can ask for payment.
When each route makes sense
This is how we would frame the decision for a subscription app that grows mainly through Meta. It is a judgement framework, not a rule, and most rows point to a test rather than a switch.
| If this describes your app | Lean towards | Why |
|---|---|---|
| Low price, and the app shows its value in the first session | Direct to store | Little to gain from a longer web journey, and the payment step costs buyers |
| You are in Apple's Small Business Program or most revenue is renewals at 15% | Direct to store | The web can lose only about a tenth of buyers before it falls behind |
| Onboarding is a quiz or personalised plan that sells the problem; higher-priced annual plan | Test a web checkout | The web suits long, persuasive journeys, and higher prices make the fee saving larger in cash |
| Most revenue is from the US or older audiences | Test a web checkout | RevenueCat finds web revenue share highest in North America and lists older audiences among the best fits for web funnels |
| iOS campaigns cannot get a timely purchase signal at your spend | Test a web checkout | A web purchase reaches Meta in the same session; trial-to-paid still arrives later |
| Store rankings and ratings drive a large share of installs | Direct to store | Web buyers do not pass through the store first, so paid spend stops feeding the listing |
| No capacity to handle tax, refunds, failed payments and account linking | Direct to store, for now | The fee saving is only real once those costs are paid |
| A direct campaign is already scaling profitably | Add web as a separate campaign | Different audiences often respond on each route; replacing a working campaign throws away a known result |
How to test a web funnel without fooling yourself
If the decision table points you towards the web, run it as an experiment with a clear finish line rather than a migration.
- Run it as a separate campaign. Keep your direct campaign running. Meta treats a change of optimisation event as a significant edit that sends an ad set back into learning, and switching a live campaign also loses the baseline you are testing against.
- Fund it to read the deepest event you can. The web campaign needs enough purchases a week to leave the learning phase. Our Meta budget guide shows how to size that.
- Judge on net revenue per pound. Take store commission off one side and payment, tax, refund and chargeback costs off the other, and compare cohorts of the same age. A web cohort with a paid introductory period and a store cohort with a free trial need to be compared after the trial has ended.
- Follow the web buyers into the app. Check how many download the app, log in and are still active a month later. A web subscriber who never opens the app is a refund or a chargeback waiting to happen.
- Recheck the rules before you scale. Store fees on link-outs changed twice in 2026 and the US rate is still undecided. Put a date in the diary to check again.
Reading those results takes the same discipline as any cohort comparison, which our subscription app ROAS benchmarks guide covers in more detail. The routing decision is one part of a broader subscription app marketing strategy.
Where we fit
We do not build web funnels, payment systems or account linking. That is product and engineering work, and an agency that offers to do it alongside your ads should be able to show you where it has done it before. What we do on our Meta Ads service is the paid side of either route: the campaign structure, the optimisation event, the cohort reporting that compares routes fairly, and the creative each one needs, built for consumer and subscription apps.
Our published pricing starts at £1,500 a month for accounts spending up to £20,000 a month on ads, with a £3,000 minimum monthly ad spend. If your app is already spending at that level and you want a clear read on which route earns more, you can apply to work with us.
Frequently asked questions
Is a web-to-app funnel better than app install ads on Meta?
Not by default. A web checkout keeps more of each payment because no store commission applies, but it usually converts fewer people at the payment step. Against Apple's 30% first-year commission, a web funnel with about 6% payment costs can lose up to roughly a quarter of its buyers and still match the store route on net revenue. Against a 15% commission, it can lose only about a tenth. Which side of that line your app lands on is an empirical question, so test it as a separate campaign rather than switching.
Does a web-to-app funnel avoid App Store and Google Play fees?
A purchase that starts from an ad and completes on your website, with no link from inside the app, carries no store commission. Apple's App Review Guidelines let an app give access to subscriptions bought on your website, provided those items are also available as in-app purchases in the app. You do take on payment processing, sales tax, refunds, chargebacks and failed payments yourself, which RevenueCat measured at about 6% of revenue for its own app.
Can US apps link to a web checkout from inside the app in 2026?
Yes. A US court injunction currently stops Apple charging any commission on purchases made through external links in US storefront apps. In December 2025 the Ninth Circuit ruled that Apple may charge some commission and sent the rate back to the district court. Apple proposed 15% for standard apps in August 2026, and that rate is not settled. Google's US external content links programme charges 10% on recurring subscriptions completed within 24 hours of an in-app link, with reporting from 1 October 2026.
How should I compare a web-to-app campaign with an app install campaign?
Compare net revenue per pound of ad spend on cohorts of the same age, after store commission on one side and payment, tax and refund costs on the other. Do not compare Meta-reported ROAS across the two, because the web campaign is measured through the Meta Pixel and Conversions API while iOS app campaigns rely on Aggregated Event Measurement or SKAdNetwork, which see different things on different timetables.
Do free trials work in web-to-app funnels?
Less well than in the app. RevenueCat notes that web buyers often enter cards with no balance, which is why many web funnels use a low-priced paid introductory period or a money-back guarantee instead. Rise Science told RevenueCat that a heavily discounted first month outperformed a free trial on its web funnel for both conversion quality and ad optimisation.
The short version
A web checkout trades buyers for margin. Work out how many buyers your store fee lets you lose, test whether the web keeps that many with ads built for the web, and judge both routes on net revenue per pound from cohorts of the same age. If you pay Apple 15%, the bar is high. If you pay 30% and your onboarding already sells the problem well, the web is worth a proper test.