A report on best practice for local data initiatives

Between June and October I worked for Power To Change UK on analysing best practice in data-driven initiatives centred around a local community. Today we publish the results.

Data for the Community

Power To Change!—Use data!

The report published today by Power To Change UK, written by Mark Braggins and me, is the culmination of several months of work during which we engaged with data initiatives around the country to understand their pains and their strengths.

The premise of the research is simple: instead of focusing on data for data’s sake, let’s support effective uses of data that can result in better outcomes for the community and increase the chances of success in funding applications.

Power To Change has this evidence-based approach to local projects in its DNA: funded by the Big Lottery Fund with a £150 million endowment to be spent over 10 years, it focusses on community businesses to create better places. When Gen Maitland Hudson, who is Head of Data and Learning at Power To Change, got in touch with me earlier in June, we decided to work together to uncover what’s happening with data at a local level, in order to help Power To Change adopt a strategy in support of data-driven initiatives that improve local communities.

Gen’s foreword

The report contains a narrative assessment of the extent to which data is being used at a local level by a cross-section of local authorities, charities, democracy activists, civil society, and community businesses. Note that when I talk about data, I’m not necessarily talking about open data — the focus of the research was on learning what is going on with data in any possible way. It felt right to expand the horizon to private data, especially data that is collected for a specific purpose. That’s why we often use the phrase data assets, rather than datasets: information is often held in unstructured forms, which includes paper surveys for small organisations. Most pieces of information will, at some point, translate into data tables, but we didn’t want to limit our understanding and engagement by speaking directly of datasets, a word that strongly associates with a specific way of storing and processing information.

You can download the report here.

Data aggregation works, if it is at the right level

What did we find? First of all, we record that there are several pockets of activity in the data space at a local level. The data used, however, vary. The most mentioned dataset is the IMD — Indices of Multiple Deprivation, which many organisations, especially those operating in the charity sector, use to build an initial understanding of their users’ demographic profile. The IMD aggregate data at the LSOA level, which seems to be just about right for many users.

A very good reason for why LSOA-level is good was offered by David King, Innovation Lead at Local Trust and previously at HACT: if you work for a housing association, a LSOA will very likely coincide with the boundaries of a single housing estate. Therefore, every data about that LSOA is in reality data about the estate’s residents. Understanding where data aggregation matches the level at which an organisation runs its operations and using data assets at that level of aggregation can power highly targeted initiatives.

Small Data? Big Data? Open Data? Closed Data?

We didn’t register many uses of Big Data. Actually, I would say there is no small data / big data question at a local level: most data is data that makes sense at a small scale, which might be derived from larger datasets. The IMD is, once again, a good example of this, together with data coming from Ordnance Survey and the Census. Such type of data is often used to understand the demographics of a population or the characteristics of a geographic area; when devoted to operations, such general data is then augmented by purposely collected data. The extra data comes from surveys — often web surveys, but we have also heard of traditional questionnaires administered on paper.

The general message is that any type of data that can improve operations is sought and used. In short, organisations just “milk the data”: they use available datasets as much as they can, and when they think they know enough, they often start collecting their own data. It is significant that many organizations are using data to inform their operations, and they need support to continue doing so effectively.

Overall, many interviewees reported uses of Open Data, and I’m clearly happy about this. In my previous report for NHS England, I was constantly asking for examples of open data uses, receiving in turn replies about closed data. For this piece of work, I asked for data in general, but the answer often was about open data.

4 steps for effective use of data

At the end of the report, we recommend 4 steps to guide organizations achieve effective data driven operations by devising a hands-on, problem-oriented data strategy:

  1. Use readily available data to understand an area and its population
  2. Collect new data to address specific questions
  3. Engage with local experts, and follow examples of local uses
  4. Develop your data framework, and seek support.

One good thing

If I have to mention one positive thing I’ve learned, above all: there is a lot of support for any organization that wants to embark on a process of using data to improve its delivery. This help comes in a variety of ways: local volunteers, paid consultancy, philantropic organisations like Data Kind. The help is not always free, but there are ways to choose an approach that suits every organisation and to adopt a data strategy that is compatible with the available resources.

One bad thing

As much as support is readily available, many organisations seem unable to share data assets and resources. This seems to be common in the charity sector. There are often data protection issues, but it would be disingenuous to think this phenomenon is always caused by statutory requirements: a highly damaging “not invented here syndrome” prevents effective data-driven collaboration to emerge.

What next?

This research was carried with Power To Change’s goals in mind, especially their need to fund local initiatives and evaluate their effectiveness. It will be interesting to see how things develop for other funding bodies. One important step for funders is, I believe, to start encouraging joint work between different organisations, similarly to academic consortia, promoting ethical data sharing that supports improved outcomes.

There needs to be investment in supporting uses of data that go beyond the mere embellishment of funding bids: data-driven operations require ongoing support, and all initiatives should be given suggestions on how to properly embed data in their processes and projects.

…and of course, I see a little lesson for the Open Data community as well — something I’ve been saying for a while: we need to shift our strategy to be outcomes- and problem-driven, even if the data doesn’t look sexy. By doing so, we will foster an environment where data is pervasive, and better Open Data will be just one of the ordinary deliverables.

The End.

If you want to hear more data-related news, please subscribe to my weekly newsletter. It’s been going for over 5 years.