Should You Send Paid Traffic to a Web Signup or an App Install?

The web-to-app question is usually framed as a fee question and decided as a funnel question. Store commission is the headline, and the handoff between paying on the web and using the app is where the money actually goes.

Both routes are defensible for a subscription app. Which one wins depends on how considered the purchase is and how much of a funnel you can afford to maintain.

Should your ads send people to a web signup or to the app store?

Send paid traffic to a web signup when the subscription is considered, the price point is meaningful and you already have a paywall that converts, and send it to the app store when the purchase is impulsive or you are still learning what converts.

The web route trades conversion rate for margin and data, and that trade only pays when the price is high enough to be worth the lost users.

A rough test: if your annual plan is $60 or more and your buyers take time to decide, the arithmetic usually favours the web. If you sell a $4.99 per month plan to people who downloaded on a whim, the extra steps cost more than the commission.

Store traffic also wins when you have nothing built. An App Store product page converts page views to installs at 8.56% on average, per AppTweak’s analysis of 2025 US data, with categories running from 5.2% in trivia games to 52.8% in food and drink. That page is already built, already localised and already tested by Apple.

For most teams the honest answer is to run both and compare on revenue per ad dollar after fees. Our piece on why app installs do not convert into trials covers the measurement discipline that makes that comparison readable.

What does a web funnel actually save in store fees?

A web funnel saves the difference between store commission and card processing, which is 15% to 30% against roughly 2% to 3%, per RevenueCat’s 2026 guide to web-to-app funnels. Which end of that band applies to you is worth getting right.

Per Apple’s and Google’s published service fees, Apple charges 30% in the first year of a subscription and 15% from year two, or 15% throughout under its Small Business Program, while Google Play has charged 15% on auto-renewing subscriptions from day one since 2022, restructured during 2026 as a 10% service fee plus 5% for using Google Play Billing, which lands at the same 15% for almost everyone.

The web side is not as cheap as the processing fee suggests. Once chargebacks, tax handling, fraud tooling and the funnel software are included, the blended cost of selling on the web lands nearer 6%, per the same RevenueCat guide.

So the realistic saving is roughly 9 to 24 points of revenue, not the full 30. On $100,000 of monthly subscription revenue, that is $9,000 to $24,000 per month, which is what has to cover the funnel’s build cost, its maintenance and the users it loses.

RevenueCat is explicit that the saving is not guaranteed, because the fee advantage can be erased by the drop-off the extra steps introduce. That is the part only your own numbers can settle.

Where does a web funnel lose money that an app install does not?

A web funnel loses money at three points an app install never touches: the page before the paywall, the card entry, and the handoff into the app. Each is a drop-out the store flow either does not have or handles with a face scan.

The published funnel numbers are sobering. Per FunnelFox’s State of Web2App 2026, which covers eight App Store categories across five regions using its own platform data, about 13% of funnel sessions reach the paywall and roughly 3% complete a purchase, and the largest drop-off sits in the quiz and onboarding steps rather than at payment.

Most of the work in a web funnel is therefore spent before the price is ever shown, which is the opposite of where teams put their optimisation effort.

Payment friction is the second leak. A card form is a harder ask than a stored store credential, and it is where international traffic falls away fastest.

The third leak is the one teams forget to measure: people who pay on the web and never activate in the app. Entitlement syncing, deep links and a sign-in that recognises the purchase all have to work first time, and a paid user who cannot get in is worse than a lost signup because you now owe them a refund.

The lifetime value is not automatically better either, and here the two datasets flatly disagree. Per Adapty’s 2026 in-app subscription benchmarks, in-app lifetime value in graphics and design apps runs about $4 higher than web over twelve months, $40.10 against $35.80, which Adapty puts down to weaker web retention after month one.

Per FunnelFox, the reverse: web lifetime value typically beats in-app across plan durations. Your own cohorts are the tiebreak, which is a reason to run the test rather than read about it.

	Web signup first	App install first
Fees on revenue	About 6% blended, per RevenueCat 2026	15% to 30% store commission
Steps before payment	Page, onboarding or quiz, card entry	Store page, install, in-app paywall
Payment friction	Card form, higher drop-off	Stored credential or Apple Pay
Attribution quality	Full session-level data and an email	Modelled postbacks on iOS
Build cost	Weeks of work plus ongoing maintenance	Already built by the store
Best fit	Considered, higher-priced subscriptions	Impulse installs, low price points

Which apps should not run a web funnel?

Apps with no subscription, apps with a low price point and apps without a converting paywall should not run a web funnel. With nothing to sell before the install, a web page is a step that buys you nothing and costs you users.

Games are the clearest structural exception. Games typically collect payment after play has started, so there is very little a player can complete on the web before downloading, and the web2app pattern mostly does not transfer.

Low-priced impulse subscriptions are the second exclusion. When the plan is a few dollars a month, the commission saved per subscriber is small and the extra friction removes more revenue than the fee does.

Early-stage apps are the third. If you have not yet found a paywall that converts in-app, building a second funnel doubles the number of things that can be wrong. Fix the in-app conversion path first, which is what our piece on turning app users into paying customers is about, and revisit the web route once the paywall works.

What changed about linking out of an iOS app in 2026?

What changed is the in-app link-out, not the right to sell on the web. Selling a subscription on your own website has always been permitted, and the restriction was on mentioning it inside the app, which is what US litigation has been about.

As of October 2026 the position in the US is that link-outs are allowed and Apple is not collecting a commission on them, following the April 2025 contempt ruling.

The Ninth Circuit upheld the core of that injunction but held a total ban on commissions too broad, ruling that Apple may charge a commission based on costs genuinely and reasonably necessary for coordinating external links and no more, and sent the question back to the district court, which has not set a figure.

The litigation is still live. Apple lost its bid for rehearing before the full Ninth Circuit in March 2026 and petitioned the Supreme Court in May 2026, and the Court granted review that June on a narrow question about the civil contempt standard, with a decision expected in 2027.

The practical consequence is that the link-out rate is unsettled and the external-purchase entitlement rules differ by region. Build the funnel so that the fee assumption is a variable you can change, not a number baked into the model.

Do not confuse the link-out with the funnel. A web-to-app funnel acquires the user on the web before they ever open the app, so it does not depend on in-app link-out rules at all, which is why it has been the more durable pattern throughout.

How do you compare a web funnel against an install campaign fairly?

Compare them on net revenue per ad dollar over a fixed window, with fees deducted on both sides, and nothing else as the primary metric. Conversion rate flatters the store route and fee savings flatter the web route, so either one alone gives you the answer you were hoping for.

Fix the window before you start. Thirty days captures the trial-to-paid transition and 90 days captures the first renewal, and a web funnel that looks better at 30 days can look worse at 90 if its subscribers renew less often.

Run the two routes concurrently on comparable audiences and creative, not sequentially. Seasonality and creative fatigue move results more than the landing destination does, and a sequential test attributes those swings to the funnel.

Measure the handoff explicitly. Paid on the web, downloaded the app, signed in successfully, and still subscribed at day 30 are four separate numbers, and the gaps between them are where a web funnel either earns its keep or quietly fails. Our guide to tracking app signups and subscriptions after iOS privacy changes covers how to keep both sides measurable.

What should you test first if you already run app install campaigns?

Test the funnel on one audience and one offer, with the same creative you already run, before you touch anything else. A web funnel is a landing-page change, and treating it as a replatform invites five simultaneous variables.

Start with your best-performing creative and your highest-intent audience, because a new funnel needs the easiest possible traffic to produce a readable result. Testing a web funnel on cold prospecting traffic at a low budget tells you nothing.

Keep the install campaign running at full budget alongside it. The comparison is the point, and pausing the control to fund the test destroys the only baseline you have.

Then budget for maintenance rather than a launch. A funnel needs price tests, quiz iterations and payment-method additions to keep improving, and the teams that win with web2app are the ones treating it as a product surface. Where the ad promise, the paywall and the store listing all have to agree, that sequencing is the work our app launch service exists to run.

The short version

Choose by price point and intent. A considered subscription at a meaningful price justifies a web funnel, and an impulse install at a low price does not. An App Store product page already converts page views to installs at 8.56% on average, per AppTweak’s May 2026 analysis of 2025 US data, and it cost you nothing to build.

Know what the fee saving really is. Per RevenueCat’s 2026 web-to-app guide, store commission runs 15% to 30% against roughly 2% to 3% for card processing and nearer 6% blended, so the realistic gain is 9 to 24 points of revenue, which then has to pay for the funnel and the users it loses.

Per FunnelFox’s State of Web2App 2026, about 13% of funnel sessions reach the paywall and roughly 3% purchase. On lifetime value the sources conflict: per Adapty’s 2026 benchmarks in-app beats web by about $4 over twelve months in graphics and design, while FunnelFox reports web ahead.

Settle that in your own cohorts rather than picking the study you prefer.

Then measure it properly. Net revenue per ad dollar after fees, over fixed 30 and 90 day windows, both routes running concurrently on comparable traffic, with the paid-to-activated handoff tracked as its own funnel.

Treat the US link-out rate as a variable while the litigation continues. Where this stops applying: a free app with nothing to sell before the install, and a game, where payment comes after play starts and there is almost nothing to complete on the web.