Quantitative testing Betagov content and layout

I am Nick Breeze, one of the Senior Insight Managers at GDS and I work in the team responsible for testing content being developed for gov.uk.

Recently we completed quantitative usability testing to compare the performance of content and layout from Directgov with that of the forthcoming beta version of gov.uk.

This follows up conventional face-to-face usability testing that was conducted in December 2011, and used GDS’s Summative Test methodology, which lets us reach a large, representative sample of 1,800 online users cheaply and quickly.

Conventional usability testing involves users being in an unfamiliar environment, using someone else’s computer, and being quizzed by a moderator. Quantitative usability testing has the added bonus of participants completing tasks in their homes, on their own computers, with no interference from a moderator.

The research involved testing how well we meet user needs in two areas:

  • Tax
  • Going to Court

Each featured a version using the existing Directgov content layout and style, and another using the Betagov version. These were then loaded up onto a test website, and participants from an online research panel were asked to take part in our study.

Screening questions, such as age and gender, were asked to ensure that we obtained a representative sample, and then each participant were tasked with finding four separate pieces of information. For example, one task asked people to imagine that they were a victim of a crime and the case was being taken to court. They were then asked to find information on who would contact them to offer advice and support. Following this they reported on how easy the information was to find, whether they had read all the content or not, what they thought of the writing style, and finally their opinion on the amount of information provided.

Overall the results showed that participants who had been asked to complete the tasks using Betagov were quicker and more successful in finding the correct information.

Some key findings from the research (attached in a slide deck below) are as follows:

  • Participants found information faster using Betagov – an average of 80 seconds compared with 123 seconds for Directgov versions. On some individual tasks it was over a minute quicker.
  • In general, we see a step change in successful task completion rates for betagov, compared to DirectGov, up to nearly 70% from around 60%. This would equate to well over 1 million more user needs being successfully met each month.
  • An average of 90% of participants saying it was ‘quite/very easy’ to find information on the Betagov versions.
  • The majority of participants described the Betagov style as ‘straightforward’, ‘to the point’, and/or ‘reassuring’. Moreover, the majority of participants also agreed that Betagov content contained the ‘right amount of information’.

At GDS we’re always very wary of making *too* much of user testing – there’s no substitute for putting a real product in front of real users with real needs. However, we can’t help but be cautiously excited about these results.

By way of context, if (and it is a big if) the same level of improvement were mirrored across the live service, a Betagov-style product could lead to:

  • Over 1 million more user needs being successfully met each month (aka a million frustrating & expensive failures avoided)
  • Users saving over 215,000 hours of their time each month
  • Quantifiably lower levels of user frustration, and higher levels of user reassurance

But to reiterate we are only cautiously optimistic: even a large 1800-strong panel is no substitute for real users facing real needs. When the beta of gov.uk goes live it will doubtless reveal many further opportunities to improve.


  1. Isn’t this a bit like A/B testing finding a needle in a haystack, vs a needle hidden amongst a few straws?

  2. It’s interesting how much of this focused on the textual content of the site (as opposed to gfx and interaction) – bring back the job description “technical writer”? Also, did you do any tests of statistical significance?

  3. Hi,

    Yes we did conduct significance testing, and all the tasks where Betagov outperformed Directgov content/layout were statistically significant. I’m sorry but i don’t understand your first question?

    Kind regards,

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s