Skip to main content

Benchmarking GOV.UK and what we are doing with what we found out

Posted by: , Posted on: - Categories: Analysis, Research stories, Team sport

It’s good practice to measure the performance of your site, and monitor this over time – often called benchmarking. On GOV.UK, we’ve just completed our fourth wave of benchmarking.

To measure performance we had a panel of representative users performing a set of common tasks that we identified through analytics data. We then focus on the poorest performing tasks in lab-based research days to get a deeper understanding of the problems.

In this post I’ll describe one of the key findings from our latest wave and the A/B/C tests we are running as a result.

Start now … or not

The most interesting finding in this wave was people’s reluctance to click on the ‘Start Now’ button on some pages. We saw this most clearly on our content on applying for an overseas passport, contacting DVLA, and calculating your child maintenance.

We kept seeing users find the correct page to complete their task, then navigate away from the page rather than clicking the Start now button. Even more interestingly was how users navigated in loops, returning to the same page many times.

During the lab days, the user’s frustration was obvious. It became clear to everyone observing that the Start now button design pattern is not always working for users.

Data helped us to understand more

Working with a performance analyst proved invaluable to understanding the scale of the problem. Google Analytics data revealed that when users first land on the overseas passport page only 68% of them click the Start now button, compared to 89% who click the button when they see the page more than 3 times (in a session).

For the 21% of users who click after the third visit, something about looping back builds their confidence compared to the first time they see the page.

The figures are significant. The 30% of users that visit the page multiple times equates to almost 52,000 users a month. So this is a problem we need to address.

The data for contacting the DVLA showed a similar pattern.

Once users found the right page, instead of clicking Start now, thousands were doing in-page searches, compared to the hundreds we see on other pages. In our lab sessions, one user landed on the correct page 6 times before eventually giving up as she didn’t think it would give her the telephone number she wanted.

She isn’t alone in this journey. The analytics data tells us that almost 1 in 3 users visit the page more than once. And in the same month we did the benchmarking, almost 31,000 people visited the page more than 4 times, with only 60% of them clicking Start now.

A/B … /C testing

After we shared these findings across GOV.UK, we created a mini team to work on improving these journeys. Or at least understanding them better.

Our hypothesis is that the words ‘start now’ do not always match users’ thinking about what they’re trying to do. The phrase suggests starting something more serious, formal or personal than they expect.

So, on 3 pages, we’re going to change the text of this button to something more appropriate to the context of what users are trying to do.

As before, we’ll measure the percentage of users who click the new ‘start now’ button, the number of repeat visits to these pages, and the task completion rates.

We hope to see a reduction in the number of times a user visits the page, an increase in them completing the task, and hopefully to reduce the number of page views involved in their entire journey.

The team have jointly identified 3 pages with start now buttons, 3 labels to test. For each page, we will do an A/B/C test with the current ‘Start now’ label, and 2 alternative labels. We will run the test for 2 weeks to see if changing the content of Start now buttons improves their performance.

Page Current label Alternative 1 Alternative 2 Start now Get application information Next Start now Calculate Estimate your child maintenance Start now Find out how Get more information

Come back for the results show

Keep an eye out for a future post where we’ll cover what we learn from the A/B/C test.

Keep in touch. Sign up to email updates from this blog. Follow Dipa on Twitter.

Feature image by the GOV.UK team.

Sharing and comments

Share this page


  1. Comment by Rosie Cowling posted on

    Hi Dipa! Are you able to share any of the findings on the start button? It would be really interesting to read as i'm currently working on the school performance tables beta service - one which isn't transactional with an end or start point. Therefore i'd like to know if people's hesitations are anything to do with the wording on the button.

    • Replies to Rosie Cowling>

      Comment by Dipa Shah posted on

      Hi Rosie. Do you mean the findings from the initial research or from the A/B/C test? We plan on writing another blog post with the outcomes of the test so keep an eye out for that. The research did uncover that part of the hesitation to click was the button wording as it didn't always match their need. However, this varied by the service as in some instances the hesitation was less the wording but more that people did not expect to have to click anything to find what they were looking for and therefore assumed they were in the wrong place. We are hoping this work will help us to better understand what we can do to encourage people to get to what they need more efficiently.

  2. Comment by Scott G posted on

    Was there any thought to make the Alt 1 option consistent across each of the three pages? (e.g. using 'Next')

    That way you could compare the results within each page (options A/B/C) and with each other (option B across 3 pages).

    My thought would be that a consistent sitewide label for moving to the next step might be useful.

    Also, have you found any correlation between simple labels (like 'Next') and increased engagement from site users that don't have English as a first language?

    • Replies to Scott G>

      Comment by Dipa Shah posted on

      Thanks Scott. Using 'Next' was one of the options discussed but we decided for this initial wave to see if different labels worked better. Depending on the outcomes of this A/B/C test it may well be one we revisit for the next phase. 'Start Now' tends to be consistent across the site and that is part of why we are doing this, as it is not always appropriate so we thought testing some relevant labels for the services might be worth exploring to see if there is any difference in the number of users clicking through. At this stage we have not thought about the differences for different users but it's a great suggestion and something that will be worth exploring when we do further research on this - so thank you!

  3. Comment by Mike posted on

    When can we expect next part with findings?
    Please share it with us!