Getting critical

At the end of August Jonathan, our User Researcher, headed down to London for two days of one-on-one interviews with representatives from the Greater London Authority, Department of Health, Institute of Fiscal Studies, DCLG, Local Government Association and the CBI to get additional insights on the Beta build.

This group was drawn from what we refer to as our ‘critical friends’ list – a group of expert, regular users of the ONS website whose requirements and expectations tend to be more specific than the ‘information foragers’ we often work with during our lab based usability testing. Often their work is reliant on ONS data so they have very clear ideas of how the current site is letting them down and what they need from any replacement.

Thankfully there was a general feeling from participants that we were moving in the right direction and they could see progress was being made. There were plenty of positive comments and an agreement that the Beta was a major improvement on the current site already – with some of the early search improvements and some design decisions coming in for particular praise.

However they are not called critical friends for no reason and a number of themes emerged that we really need to crack before the site would be ready to lose the ‘Beta’ warnings – here are just a few of the things now being added to our backlog;

  • Methodology needs to be more comprehensive and include overarching methodology information and historical versions of specific methodology. This is one of the risks of taking such a ‘data driven’ approach to our content planning – we missed how important some content was to people even though it got minimal traffic
  • They did not want to have to know where in the taxonomy to find things. At the moment browse really relies on the statistical taxonomy being intuitive – research demonstrates it is but many users are getting frustrated by how many levels they need to travel and understand. We are investigating solutions to this – in particular adding cross-cutting content to initial browse pages will hopefully help
  • Search, while much improved, needs to handle broad queries around data much better. Our initial solution required a bit too much up-front knowledge of the taxonomy (again). This is going to be extremely important given we know how important direct access to data is for so many users
  • Dataset pages need more metadata to be able to make decisions on whether relevant or not – a preview of the spreadsheet would be huge for them and we need to offer a way to download time series data in bulk
  • Real concerns expressed around our plans to use the National Archives web archive rather than migrate everything. This seems to be a perception issue rather than specific concerns about the TNA – we need to work out a way to allay these fears before we launch though
  • Making the connections between underlying ‘surveys’ and statistical outputs was mentioned a few times. I’ve always thought this seemed like an important offering and it is something we are already working on.

It was an intensive couple of days for Jonathan but his pain was our gain as these insights are invaluable and nicely compliment the work we have been doing around testing scenarios in the lab(s). We have more sessions with critical friends scheduled in a month or so, more lab based testing being organised, a first attempt at online un-facilitated testing on the Beta and next week we are attending the Royal Statistical Society conference in Exeter solely to speak to users and undertake some guerrilla user testing so I’m confident that we are living up to our ‘users first’ principle.