The Environment Agency (EA)'s experience going open has raised some useful lessons that other organisations considering opening their data can learn from.
Consider data licensing and terms from the very beginning
When different organisations are involved in collecting data, it is important to make sure that their licensing accounts for the data being released, whether immediately or later on.
The release of the Risk of Flooding from Rivers and Sea (RoFRS) dataset as open data was not considered during the initial agreements made with all third-party data suppliers, who may have held intellectual property rights or contractual restrictions over data included in the dataset. These agreements had to be checked and addressed in a short space of time before the data was released.
We need to get the data management process right […] so when you come to publish something it's just downloadable – anyone can use it to do what they want because we've checked all the rights are there […] We need to build [this] into how the data management function works as a whole, rather than [do] what we've done in the past, which is pick it all up at the end.
– Mike Rose, Open Data Manager, Evidence Directorate, EA
Publish your data in accessible formats
Consider different format options in order to make sure your open data is easy to access and use. Once EA’s flood data was released, a number of users complained that it was made available in a proprietary format that made it difficult to use.
Angharad Stone, of the Data, Mapping, Modelling & Information team, suggested that:
To begin with, we only made RoFRS available as an Esri geodatabase, which is a proprietary format which can only be used with certain software. So we have had people coming back and saying 'you've produced this as open data but you've put it out there in a format most of us can’t use.'
Shortly after, EA responded to the feedback from its users and made the data available in different formats. Some EA staff also suggested that their teams would benefit from specialist training on the latest technical aspects of open data publishing, such as linked and real-time data.
Create user feedback loops to evaluate your data’s impacts
Once data is released, it helps to have systems in place to monitor, assess and evaluate its impact. Without these, it is impossible to track who is using your data and for what purposes, because open data does not require people to sign-up or register to use it.
With limited resources available, EA did not put processes in place to actively engage its data users, so found it hard to evaluate its data's use and benefits.
On the one hand having no registration means that there's no barrier to entry, but on the other hand you've got no way of either helping people [to] make better use of the data or of working out the difference between someone making genuine heavy use of the APIs and someone with rogue code who doesn't realise the load they are accidentally placing on the service. So, for managing a service, [releasing open data] can be quite a hard balance [to strike].
— Dave Reynolds, CTO, Epimorphics
Tell businesses about your data
Opening data is the first step towards making it useful. Promoting your data and engaging users is key.
A number of the EA team described how they felt they didn’t do enough to ‘sell’ their open data. Guiding users to the data, describing to them what it can be used to understand and helping them to use it were all areas in which EA staff think more could be done. In particular, by making more businesses aware of the data it could enable them to make further substantial cost savings.
As Mike Rose asks:
Is the commercial companies’ [who provide flood modelling] data better or worse than ours? Who knows. But should we be putting effort into selling our free product? Absolutely. We believe it's the best stuff and we are making regulatory decisions based on it. We believe that people out there should be using it to make their decisions. If we're not telling the insurers that our data’s available for nothing, they'll only see the flyers from the companies that are going to charge them £500,000 a year."