Feedback : Search bar back in the top nav

What has changed?

The search has been moved from the sidebar (top left) to the center of the top navigation.

The hypothesis is this change improves navigation efficiency by placing the search function in a more intuitive location. The update is now live and is expected to provide a more streamlined user experience. See section on how we plan to measure success.

Note: This is currently behind feature flag :search_button_top_right and is currently scoped to GitLab team members

Desktop view

Before Screenshot_2025-03-27_at_11.07.00_AM
After Screenshot_2025-03-27_at_11.07.35_AM

Small screens

Before After
Screenshot_2025-03-27_at_11.09.12_AM Screenshot_2025-03-27_at_11.07.54_AM

Mobile

Before After
Screenshot_2025-03-27_at_11.09.00_AM Screenshot_2025-03-27_at_11.08.25_AM

Feedback requested

  • TBA

How long are we going to keep this feature flag on, and how will we measure success?

As noted in #480341 (comment 2127437875), the plan discussed with Design was:

Give this as try for internal users for 60 days and see how the metrics looked on a rolling 30 days basis during that time. If we're seeing meaningful improvement, even if those metrics just improve and stabilize, we would then gradually roll it out to a subset of users on dot-com.

Do we need a certain quantity of the user base to be exposed for statistical significance?

Yes. I did some calculating and came up with a sample size of 334, at 167 per group (assuming two groups). That should give the test a power of .85, p = 0.05

Is 30 days long enough?

It's probably the minimum, tbh, but I would expect to see some effect in that time period. Maybe 30 days is the cutoff if we're not seeing any effect, but if we're seeing good results, we keep running it.

What is the hypothesized impact to an increase or decrease to this metric?

My hypothesis is that an increase in these metrics indicates that users are again able to search, which, extrapolated out to all our users, should increase our CSAT score and improve subjective research findings about users' ability to find what they're looking for on GitLab. You probably know more than me about how our CSAT is calculated and if my hypothesis for impact there is a reasonable one.

Known bugs

  • When the search input is presented only as an icon button it has no focus indicator for keyboard users and no text that will explain its purpose to screen reader users (#527592 (comment 2420664760))
  • Search can overlap the breadcrumb depending on the breadcrumb length and the viewport width (#527592 (comment 2420608599))

Links

Edited by Ben Venker