Reflecting on our first ‘data hack’

Sheffield Data for Good has been around for about 15 months now. Its aim is to bring together the data and social expertise in Sheffield to help solve the city’s social problems. We’ve held a few Meetups, some focusing on homelessness and others on ‘social isolation’. For most of that time, the group has been finding its feet, particularly in relation to the most appropriate format for our sessions.

One Saturday in late January 2019, however, it felt like it came of age. We held our first ‘data hack’ in partnership with Roundabout, a youth homelessness charity in Sheffield. The plan I had originally envisaged back in November 2017 while speaking with colleagues at Good Things Foundation seemed to make more sense. Here are some of my (non-technical) reflections since then, as well as a few things I’ve learned in the process…

Find a space. Communicate a decent, relatively contained brief — Roundabout’s data, in this example. Make it ‘feel’ relaxed and inclusive. And people actually show up (with snacks too!). At 10am on a Saturday. For 6 hours. I know; it’s remarkable, isn’t it?

This shocked me a little actually. It bowled me over with excitement and enthusiasm too. Although I believe wholeheartedly in what we’re trying to achieve as a group of people, I did find myself asking this question: if I wasn’t organising it, would I have actually turned up? I’m not sure I’ve reached a conclusion on that yet.

People are pretty amazing. The skills that people are willing to share voluntarily are highly technical. Some clever things went on in that room. And no-one was there for personal gain or to say ‘look at me!’. The prevailing culture on the day was one of generosity, collaboration and mutual support/learning.

Here are some of my ‘recommendations’ for holding such events

To understand better what motivates people to come along, ask them. And ask everyone. Individually. Face to face.

I think I could have done this part better. From the few conversations I did have, here are few of the things people said, or at least my interpretation of what was said…

  • I know I have a skill set. Applying it in my day job isn’t enough. I want to use it for ‘social purpose’.
  • This is actually a fun Saturday for me.
  • I want to learn from other people in relation to technical skills
  • I’m not a data expert but I want to learn about different ways to unpick a problem in a more general sense.
  • I want to build my personal confidence.
  • My employer is encouraging me to link up with the local scene.
  • In my sector, I’m not expected to produce an early output so quickly. It’s refreshing/exciting to have an opportunity to try things out without being too precious.

In January’s case, this was Amy from Roundabout. Amy knew what the data actually meant. She was an essential reminder that there were people behind the spreadsheet rows. There were nuances to what a particular column meant. And all of this had a real implication for the people supported by Roundabout, as well as those working for the charity.

Having this understanding in the room pegged the conversations in the reality of what actually happened on the ground. It was also fundamental in shaping the way people analysed, visualised and computed things. People wanted the outputs to be useful; to actually inform what Roundabout did and highlight potential areas for change.

As Dan Olner’s excellent blog on our January hack points out, it’s easy to assume that complicated data magic is what people are after. In reality (and as an example), a simple calculation of the average number of ‘interactions’ with Roundabout per person can be hugely enlightening for those collecting the data. Often the barrier to data analysis is not capability in the charity/voluntary sector; it’s the time and resource required to do it.

I was so chuffed that people turned up. The format seemed to be working. People were producing things. People were talking to and learning from each other. It was all going swimmingly.

It did actually go well. On reflection though, the social element of the format probably doesn’t work for everyone. The same probably applies to showing/sharing your work in a public document as some people may find this daunting. Some people simply prefer to work on the fringes of things without the pressure to ‘produce’. Others may prefer a little more support with how they work in groups.

I think we could have been more sensitive to these needs. I very much hope that our next hack day (the 9th March 2019) will factor these in much better. You’re all welcome to come along, by the way.

It would be easy to squirrel away in silos to do our data thing. In the spirit of sharing little and often though, we encouraged people to dump their outputs — graphs, sentences, links, code, photographs — in to a shared document as we went. We used Dropbox Paper for this. Furthermore, we came together every hour as a whole group to talk through what was new on the shared document. People had the chance to say ‘I did that and it means…’ but also ask if anyone else has done anything that might fill in a gap for them.

There was also a good amount of analogue content produced. The pen is not dead. And drawings are very useful.

There were two people I didn’t expect to turn up (one of which was my dad, but that’s not the unexpected bit). Neither considered themselves as data experts (or even data analysis beginners). Neither considered themselves as homelessness experts either. They were interested in how the Meetup was organised and formatted as a way to understand a challenge more generally. This meant that they walked around a lot asking questions. What tools are you using? What does that diagram mean? Please can you explain that to me in words? Have you seen what so and so has done over there?

Although unexpected, this really joined things up. It provided another perspective. In turn, this supported people to reflect on the ‘usefulness’ of what they were producing and to think about the best way to communicate what it all meant.

About 30 minutes in to the day, as we dispersed from the introduction/brief session, there was a palpable sense that some people felt the brief was too loose. Some people are more comfortable with this than others.

Where do I start with this? What are the questions we’re trying to answer? Shouldn’t we work in groups formed around our skill sets? These are all valid questions but I think we could have done more to say that this uncertainty is OK (and even to be encouraged) at this point.

On reflection, this could have turned out badly if we had wasted the first hour of doing time because everyone was unsure what to do. Most of the uncertainty dissipated pretty quickly, however, as soon as the first graph was publicly shared. It set the tone and provided a starting point for some people to get going.

This was the first time I had arranged a data hack. In fact, the one I was organising would be the first time I’d ever been to one. I’d heard and read a bit about how they work, particularly DataKind’s data dives with organisations, but I wouldn’t exactly say I was confident in how to do it.

When we set up the date for the data hack, a number of people came forward to volunteer their support with organising the day. We were honest that we weren’t sure where to start and people shared their time and experiences. Particular thanks go to Sarah Miller (Jet2), Lauren Quinn (Good Things Foundation), Dean Robinson (Shelter) and Jag Goraya (Better with Data).

Although it felt like January’s hack settled on something productive, I don’t think we’ll ever stop evolving. Tweak, reflect, tweak, reflect…

I’m always up for a chat to talk about this (or anything else) so please feel free to get in touch. I’m on Twitter here.

Freelancer. Data and insight strategy through open working. Always social purpose and having conversations. Founded Sheffield Data for Good and Data for Action.