AMP

Making Your Wait a Little More Great: New Loading Indicators in AMP

Websites

This post is from the AMP Design Team. We’re product designers and researchers responsible for making sure AMP’s components and overall experience are usable, accessible, and elegant. You can find us hanging out with the UI and Accessibility Working Group on GitHub or in the Contributor Slack.

One of the biggest ways that AMP improves web users’ daily experience is by helping developers make sites that are fast. A unique advantage of AMP is that it’s always improving behind the scenes – we can roll out speed improvements to sites that use AMP without developers needing to do anything.

Performance best practices have always been baked into every corner of AMP, but actual speed is only half the story. While engineers continue to sweat the technical details, on the design team we set out to improve perceived speed. Waiting will always be a part of life – there are some parts of AMP pages our engineers can’t make faster because they come from a service we don’t control (like a Facebook post or YouTube video) or just because they’re large. But there’s plenty of evidence that the environment you’re in affects how you feel about having to wait.

If you’ve used more than a few AMP pages, you may be familiar with the loading indicator we’ve used since AMP launched in 2015:

Today we’re beginning to roll out a new one:

Let’s look at how we got here. We started with three goals:

  1. Make loading feel faster.
  2. Let readers know what’s coming. Since 2016, we’ve used a different loading indicator for ads on AMP pages so people could immediately tell them apart from the main content. We wanted to extend that to more types of content – videos, tweets, Facebook posts, and more – so you don’t have to wait for a video to load only to realize you’re not in a situation where you can watch one, for example.
  3. Modernize the design. We can’t put numbers on this, but many folks on the AMP team felt our loaders could be more visually elegant.

We also needed to keep the design elegant, but neutral. UI that ships with AMP needs to fit in well with the design of any AMP page – that means no dramatic colors or noticeable styling.

We did some prototyping and asked a panel of 2500 web users to look at one of ten different loading indicator designs that “loaded” content after a set amount of time and predict how long it took. The accuracy of the predictions wasn’t important—humans aren’t good at estimating very short durations—but the trends were. Participants estimated that they were waiting nearly a second less with some of our new designs compared to traditional options like the existing AMP loading indicator or a basic spinner. Based on this data and some observations we got excited about during the design process, we chose a design.

We landed on three principles:

  1. Sometimes no loader is better. Showing a loader before you’ve even noticed you’re waiting can be distracting and make the site seem slower.
  2. Keep it interesting. Having to wait isn’t helped by seeing the same repetitive spinner you’ve seen a million times.
  3. Keep it consistent. Having multiple things loading on one page, each with its own loader design and timing, is distracting and sloppy.

To satisfy the last principle, we made sure our design could be adapted to different sizes, content-type icons, and backgrounds. To satisfy the first two, our loader animated in several stages: first we would show nothing, then we’d show an intermediate animation, then finally we’d finish on a spinner that looped until the content was loaded. People wouldn’t see a loader at all if the content took less than a half second to load, and they wouldn’t see a repetitive spinner unless the content took a full 3.5 seconds.

Our design met all three of our goals: we had encouraging data on its perceived speed, we felt strongly that it was more elegant, and it did a good job letting readers know what was coming.

There were some additional details to figure out to maintain AMP’s flexibility for developers. We didn’t want the new design to conflict with AMP’s ability to show a placeholder image, message, or even a custom loading indicator before content has loaded. For content with a simple placeholder image, we still want to indicate to readers that something more is coming, so we’ll show a special version of the loading indicator that will stand out on any color image. For any placeholders more complex than that – messages or custom loading indicators, we won’t show the default loader at all to make sure our design doesn’t clash with yours.

The final proof will be in the live experiment we’ll be running in the upcoming weeks: launching our new loading indicators to a small percentage of AMP pages and keeping a close eye on site performance metrics as well as community feedback. Based on this feedback, our aim is to roll out our new loaders to all components on all AMP pages.

We hope you enjoy the new design and support us in making the user experience of the web better, one small detail at a time.

Posted by Andrew Watterson, Product Designer for the AMP Project at Google