For most products, mobile web is the front door. Someone searches, taps a result, and lands on your site on a phone. The app — the thing with notifications and retention — is one step away, and that step is where the funnel leaks.
Ask first whether the app is actually better
The honest starting question. If your mobile site does the job, pushing people to install adds friction for no gain.
The app is worth pushing when it offers something the web genuinely cannot: push notifications, offline access, camera or location features, or a faster experience for repeat use.
If you cannot name the advantage, the install prompt is just a tax on your web experience.
What works
The native smart banner. iOS has one built in, added with a single meta tag. It sits above your content, does not block it, and users recognise it.
Contextual prompts. After someone does something the app is better at. Not on arrival, before they know what you are.
A routed link on every store CTA. Every "get the app" button should go through a link that detects the device rather than hard-coding one store.
What backfires
Full-screen interstitials on arrival. They block the content someone came for, and Google has published guidance discouraging intrusive interstitials on mobile.
Repeat prompts. If someone dismissed it, remember that.
Prompting mid-task. Checkout, form completion, reading an article they clicked from search. Let them finish.
The install-state problem
You cannot reliably tell from a browser whether your app is installed. Attempts to detect it are unreliable and get less reliable as platforms tighten privacy.
The pattern that works regardless: attempt the deep link, and if nothing happens within a short window, fall back to the store. Installed users get the app. Everyone else gets the store. No detection required.
Measuring it honestly
The metric that matters is not banner impressions or even installs. It is whether people who came through web-to-app stay.
An install driven by an aggressive interstitial from someone who wanted to read one article is not a win. It is an uninstall in three days and a notification permission you will never get back.
Frequently asked questions
- What is web-to-app?
- The practice of moving someone browsing your mobile site into your native app, either by installing it or by opening it if they already have it. It matters because mobile web is usually where people arrive first.
- Are app install banners effective?
- The native smart banner on iOS is unobtrusive and worth having. Full-screen interstitials that block content tend to work against you, and Google has published guidance discouraging intrusive interstitials on mobile.
- Should I send every web visitor to the app?
- No. Someone mid-checkout does not want an install prompt. Target the moments where the app is genuinely better, like repeat use, notifications, or offline access.
- How do I know if the app is already installed?
- You cannot reliably detect it from the browser. The robust pattern is to attempt the deep link and fall back to the store after a short timeout, which works whether or not the app is present.
Read more
Growth · 1 min read
Referral links that actually work on mobile
Most referral programs leak at the same point: the invited friend installs and lands on a generic home screen with no idea who invited them. Here is the plumbing that keeps a referral intact.
Growth · 3 min read
One link in bio for iOS and Android: stop splitting your traffic
Social bios allow one link, but apps ship on two stores. Here's why link-in-bio pages add a step that costs installs, and how a single routing link solves it without an intermediate page.
Growth · 3 min read
App install links for creator and influencer campaigns
Give every creator the same link and you cannot tell who drove what. Give each a different link and you have a maintenance problem. Here's the setup that gets clean per-creator attribution without either.