A few UX changes. A single page. Millions of new downloads.
We recently designed, tested, and released a new version of our our primary download page for Firefox for Desktop. In our tests, we improved the download conversion rate of the top 3 non-Firefox browsers by over 12%! This alone results in millions of additional downloads annually.
Focusing on the entire funnel leading up to a product download and not just the product itself, is as important as the efforts taken to improve retention of a product. This is one of the approaches that the Websites team at Mozilla is taking to improve and support our products.
What is the /new page on Mozilla.org?
This is where the majority of our desktop browser downloads are initiated. For instance, if the user searches for "Firefox", "Mozilla Firefox", or "download Firefox" from a desktop browser, the /new URL is the top search result.
What did we change?
Though a relatively simple page, we were able to make number of changes across visual design, interaction, technical improvements and overall user experience that had substantial results. How did we do this?
1. reduced the number of steps to download
2. simplified number of actions displayed on page and reduced distractions to funnel user directly to download (ie: no link to Fx for Android or other products displayed)
3. focus on our product - the last page design did not display visuals that focused on the product, but focused more on Mozilla community.
4. significantly faster page load time
5. updated style to our responsive Mozilla style guide
6. inline page interaction that responds immediately to the user's request for download, resulting in no page refresh for confirmation and installation instructions.
Old experience - includes page refresh:
New experience - inline interaction, no page refresh:
Design decisions & Testing
We could infer that the outcome of changes such as updating this page to a consistent style as our other Mozilla.org pages and improving page load time would have a positive result. However, some of the other changes were more subjective and required testing and validation before releasing to 100% of our users.
Based on common interaction patterns, what we know about how users respond to pages that "feel" responsive to their actions, as well as minimizing distraction, we were able to make many initial design decisions. To validate the more subjective changes, such as button placement, button style, and animations, we ran A/B tests using Google Analytics Content Experiments.
An important part of testing is not just validating our work, but
exposing interesting facts about our users and issues that may need
further attention. For instance, we learned that large percentage of
users downloading Firefox already have Firefox. This could mean that the
user is not aware that we run silent updates, that they are not aware that their version of Firefox is
already up to date, that they wanted a fresh copy of the browser, or a
number of other possibilities. This is just one of the interesting things we learned that we are looking into for further improvements.
Given the success of these tests, we were very confident in releasing the new experience to all of our users. Our initial improvements to the /new page and download funnel are just the beginning. We will also continue to test improvement possibilities, such as the style of the download button, to learn more about our users and improve the performance of the download funnel.
A few stats:
Across the top 3 browsers, we saw a 14% download conversion improvement.
Across all browsers and operating systems, we saw an average of a 4% improvement.
Considerable improvements to page load time:
- Time to be able to interact: 46% decrease
- Time to load page content: 71% decrease
[ To see more stats, and learn about our A/B testing process, stay tuned for a post from Chris More. ]
Testing download button designs / styles, translate page for non-english languages, add logic and conditional messaging for all platforms, improve other touch points within the onboarding funnel by using a similar process of testing and validating with real users.