Integrate AMP with your app
This guide is intended for mobile and web app developers who want to integrate and link to AMP pages. For example, consider a mobile chat app that loads the AMP version of a shared URL to achieve a faster experience for users.
Transform links to AMP
With AMP, it's possible to near-instantly render external websites inside your native or mobile web app. You can achieve this by matching URLs in your content to their corresponding AMP URLs (if one exists) and by opening the AMP version instead of the original version. You can use tools like, Google's AMP URL API to help you with this.
For example, the following message can be transformed to serve the AMP versions by replacing all URLs with their matching AMP versions (if they exist). To reduce load time and guarantee that valid AMP is served, you should link to the cached AMP pages in the AMP Cache.
This is a message with links to an <a href="http://www.example.org/a">
article with AMP version</a> and an <a href="http://www.example.org/b"> article without AMP version</a>.
This is a message with links to an <a href="https://www-example-org.cdn.ampproject.org/c/www.example.org/a">
article with AMP version</a> and an <a href="www.example.org/b"> article without AMP version</a>.
Ways to transform links
There are three ways to programmatically transform links:
- Write-time server-side (preferred): Retrieve the AMP URL via Google's AMP URL API at write time of a URL and store AMP URLs server-side. Pass both URLs to the client because the original URL might be needed for sharing. This is the recommended approach because there are fewer client-side network requests. When taking this approach, it's important to regularly (e.g., daily) scan links for AMP versions because websites are increasingly adopting the AMP format.
- Read-time server-side (some use): Retrieve the AMP URL via Google's AMP URL API before passing the content to your client. As mentioned above, pass both URLs (AMP and non-AMP) to the client because the original URL might be needed for sharing. This method can be good for low-fan-out services.
- Client-side (if server-side isn't possible): Retrieve the AMP URL via Google's AMP URL API from the client. Use this approach if server-side URL transformation is not possible (for example, for messaging apps using end-to-end encryption). Make sure to trigger URL transformation as soon as the content is available, before any user interaction has taken place.
Google's AMP URL API
Google provides the AMP URL API to retrieve the matching AMP HTML URLs for a given list of URLs (official documentation / demo. The URLs do not need to be the canonical versions. If an AMP version exists, the response includes the original AMP URL and the URL for the cached AMP page on the Google AMP Cache.
For example, for a given list of URLs:
The response body contains the AMP URL mapping in JSON format:
"errorMessage": "AMP URL not found.",
Using AMP Caches
An AMP Cache is a proxy-based content delivery network (CDN) for delivering valid AMP documents. AMP Caches are designed to:
- Serve only valid AMP pages.
- Allow AMP pages to be preloaded efficiently and safely.
- Perform additional user-beneficial performance optimizations to content.
Currently, there are two AMP Cache providers:
This gives two choices to display an AMP file in an app by using either:
- the version hosted by the publisher
- the version hosted in an AMP Cache
We recommend using the AMP Cache for the following reasons:
- Better user experience due to faster load time and low latency (>1s faster loading time).
- Performance and bandwidth benefits due to additional caching of client dependent artifacts, e.g. caching different versions of the same image depending on the client's viewport size.
- The original AMP file might no longer be valid AMP, which could lead to a bad user experience. In this case, the AMP Cache serves the last valid version of the AMP file.
- A not-so-upstanding publisher could serve two different documents to an AMP Cache crawler and to your users. Using an AMP Cache guarantees that users always see the same AMP file as the Cache.
Implementing an AMP Viewer
The AMP Runtime provides a Viewer API, which provides a protocol to send and receive messages between the AMP Runtime and the Viewer. This makes it possible to control the pre-rendering of AMP documents, swiping between articles, and AMP Runtime instrumentation. You can learn more about the AMP Viewer API in the Connecting AMP Viewers with AMP pages guide. Viewer implementations for web and iOS are available on GitHub. An Android viewer is not yet available, see this answer on Stack Overflow for how to best configure a WebView for displaying AMP pages.
Here are some general best practices for implementing an AMP Viewer:
- Serve the AMP page from an AMP Cache (>1s faster loading time).
- Display the article's publisher origin (e.g., in a collapsible header).
- Provide a sharing action (see also the "Sharing AMP Content" section below).
- In webView-based viewers, enable third-party cookies.
- Set a referrer for your platform/app.
When sharing an AMP document from within a platform's AMP Viewer, the platform should share the canonical URL when technically possible. For example, if the platform provides a share button, this button should share the canonical URL.
The philosophy of the AMP Project is that platforms should get to choose which version of a document to present to the user. For this reason, it makes most sense to share the canonical version (as opposed to the AMP version) when sharing to a different platform, and then expect the target platform to make the right choice.