Anyone familiar with the current ONS website, particularly if you have tried to use the search will be well aware of just how much information is available and updated regularly. Identifying the content required for the alpha prototype from this huge resource was always going to be tricky and as the member of the team with the most experience of ONS outputs and the current website (having worked in publishing on both the current and the pre-2011 site for about 6 years) I was tasked with leading this work.
This meant that for a significant period of time I became a genuine user of the current site and can honestly say I can identify with many of the frustrations our users have.
Proving the point
Deciding how we would populate the site, and the level of content required for the proposed structure was the first challenge. Our aim was to make sure that there was enough in there to validate or disprove the assumptions and decisions that we had made along the way, but with a relatively small window we had to be realistic in what could be achieved in the time.
We looked at a number of options, including taking all of the content from a select few areas, however we eventually decided that the best approach to test the search and navigation journeys would be include a thinner, wider spread of information across all of the data and analysis we produce. What we decided this meant in reality was getting enough content to populate all of the pages in our taxonomy navigation.
What content do we need?
Once we had an idea of what we need to populate, the next challenge became working out which elements of the current site we would recreate on the prototype. We decided that keeping it in sync with the current site would not be feasible and we would instead deliberately go for older data to try to discourage anyone from taking potentially incorrect figures from the prototype (we also put a big red box on the homepage ‘warning’ users about the data). This left us with about 60 pages (like http://alpha.ons.gov.uk/#!/economy/inflationandpriceindices), designed to provide an overview of a topic to get content for as well as the myriad of content required for the links off these pages.
The first step was to get a feel for how much content this approach would mean we had. Starting with the taxonomy pages I went through and identified which individual time series, datasets (defined as groupings of time series or other groups of data) and statistical bulletins we would need. Even being an expert in the current site, how it is structured and the content on it, this was not a task I fancy repeating in a hurry! However in the end I had created a number of spreadsheets outlining what we needed, such as this one.
I was able to get a head start on this before the team came in to start building so by the time they started we had a good idea of what we wanted to include as well as the scale of it. At this stage we also did not know what format the content would need to be in to actually feed into the website and I must admit I did not expect us to use the files themselves as the basis for generating the JSON that the site is created from. Of course, I wasn’t disappointed by this!
I think the story of how we got the data side of things together and into the site is one for a later blog post, but I will just wrap up with a few stats (seems appropriate) about how much content we were able to add in the time;
- 52 statistical bulletins
- 84 datasets, and
- 36,653 individual time series
Perhaps a little more than was strictly required…