The Complete Guide to Deep Linking

What Is Deep Linking?

Deep linking is a technique that sends users directly to specific content inside a mobile app instead of the app's home screen. When a user taps a deep link, they land on the exact screen you intended — a product page, an article, a profile, or any in-app destination.

There are three types of deep links:

• Standard deep links — Custom URL schemes (e.g., myapp://product/123). Only work if the app is already installed.
• Universal links (iOS) / App Links (Android) — HTTPS URLs that open the app if installed, or fall back to the web. The modern standard.
• Deferred deep links — Preserve the link context through an app install. The user taps a link, installs the app from the store, and still lands on the right content after first launch.

How Universal Links Work

Universal links use an association file hosted on your domain to prove ownership:

{
  "applinks": {
    "apps": [],
    "details": [
      {
        "appID": "TEAM_ID.com.example.app",
        "paths": ["/product/*", "/invite/*"]
      }
    ]
  }
}

iOS downloads this file (.well-known/apple-app-site-association) when the app is installed. When a user taps a matching URL, iOS opens the app directly instead of Safari.

Android App Links work similarly with a assetlinks.json file and intent-filter declarations in your manifest.

Deferred Deep Links

Deferred deep links are the most powerful — and most complex — type. They solve the "install gap" problem: what happens when a user taps a link but doesn't have the app yet?

The flow works like this:

1. User taps a link on the web
2. The link service records the click context (destination, UTM params, device fingerprint)
3. User is redirected to the app store
4. User installs and opens the app
5. The SDK queries the link service with device signals
6. The service matches the install to the original click
7. The app navigates to the intended content

WarpLink handles this entire flow with a single SDK method call. The matching uses a cascade strategy: referrer, device ID (IDFV), then probabilistic fingerprint matching.