FAQ

Is quicklink guaranteed to make your web pages load faster?

No 😄quicklink attempts to use a heuristic (pages in the user's viewport) to decide what links to prefetch for an experience. These may be pages your user is going to visit next, but there is no guarantee this is the case. With this in mind, we recommend measuring the impact quicklink can have on your experience (tools not rules and all that.. :))

WebPageTest supports measuring page-load performance from one page to another (via it's support for scripting and this is the mechanism we use to determine if quicklink improved performance for an experience.

For example:

navigate    https://example.com
sleep    30
navigate    https://example.com/page-2.html
sleep    30
navigate    https://example.com/page-3.html

A baseline test might measure how long it takes to load and navigate from the homepage -> page-2 -> page-3. A comparison might look at whether quicklink improved how quickly page-2 or page-3 could load up when prefetched.

What kind of performance improvements can I expect to see if I use quicklink?

This can vary based on the architecture of your site and links within the viewport users actually end up navigating to. That said, we can look at a case study of using quicklink on some leaf pages for a re-hosted version of some sites for science.

We cloned some pages from The Guardian, re-hosting them on Firebase. For the same user-flows (navigating between three pages), a page prefetched with quicklink was able to fully load on 3G network connection/Moto G4 in just 1.3s (bottom row). Compare that to the 1.9s it took without any prefetching (top row).

WebPageTest traces for the before and after are both available.

We similarly cloned some pages from the Google blog (blog.google). For a user-flow from the homepage to a new story, quicklink only shaved off 0.1s when this page was prefetched. This is one demonstration that prefetching isn't a silver bullet and while it may have been worth using in the case of the Guardian, probably makes less sense here.

This is in no way exhaustive benchmarking. We hope to do more of this in the near-future.

Does quicklink improve the first-load experience of web pages?

quicklink is focused on attempting to improve the user-experience of subsequent navigations on a site. This means that it won't improve page load performance for the first page in a funnel loading it, but it should hopefully improve perceived latency of the next page (if it is a page that met the criteria for being prefetched).

Doesn't eagerly prefetching pages waste bandwidth and system resources?

quicklink intentionally checks if a user has opted into a Data Saver mode or is on a network connection that is effectively slow. If this is the case, we bail on prefetching entirely. We want to be mindful of your data usage. If however you're on a connection that is effectively faster (think 4G, WiFi etc) and you haven't turned on a Data Saver mode for your browser, we do prefetch.

Shouldn't browsers explore automatically using some of these types of optimizations?

Some browsers have (but don't necessarily use the same heuristics quicklink does). Chrome has supported opting into a Use a prediction service to load pages more quickly mode.