The data landscape is ever-evolving - so how do public institutions manage? We discuss with Parliament's former Head of Data & Search, Dan Barrett.
Hi! 👋 Who are you?
I’m Dan Barrett, the new Head of Data Science at Citizens Advice.
Tell us a bit about your job.
It’s a new role in Citizens Advice, focusing on establishing data science in the organisation. I will be improving data capabilities with a view to helping Citizens Advice meet its users' needs even better than it does at the moment.
You used to work as Head of Data & Search at Parliament. What was that like?
In terms of remit and responsibilities, that job was similar to the one I have now. I built from scratch the first data specialist team in the UK Parliament’s Digital Service.The team improved the quality and discoverability of data provided by the organisation, developed a data strategy for Parliament, and built a new open data service.
"I think there’s a lack of definitive facts in the online ecosystem"
How has the data landscape changed over time? Are institutions struggling to keep up?
Expectations have changed in the last 20 years. I believe that people expect concrete answers - the outlook is less that you’re going to browse through a lot of web results to find an answer, and more that the definitive piece of advice will be provided to you.
I think there’s a lack of definitive facts in the online ecosystem. As a small but repeatable example, during the pound coin replacement, I searched: “what do I do with my old coins?" I expected a conclusive answer from the Treasury or Royal Mint. But I was surprised to see that the majority of results were from news outlets and media companies.
Institutions are in an environment where they are good at publishing but not as efficient with contextualising and explaining that material for reuse by machines and citizens. It’s vital to ensure there’s information surrounding the data so it can be understood. However, providing that information is quite complicated to achieve because institutions need to know how to optimise material to perform well on the web.
Over the course of your time there, the data landscape must’ve changed a lot … what were the challenges you faced with the changes?
The biggest challenge was understanding what the users needed and being able to meet those needs. One of my main frustrations was that we didn’t get as far as I would’ve liked in really transforming how data is published on a platform level.
The challenges aren’t about one individual or team’s efforts; they require backing from institutions. There were a few good examples from other organisations that really made me think about this, but in my own case I wasn’t in a position where I was going to make a significant breakthrough in those lines.
"The absolute basics towards a perfect data world would include structured data that is easily accessible, discoverable, contextualised, understandable, definitive, and well-designed"
Chances are for government institutions, or at least for multiple Parliaments that I know, you wouldn’t have had the same systems for all these years. The information needed to be implemented so data can be collated over a course of time - and that’s a huge challenge.
What would a perfect world look like, from a data perspective, for citizens? What steps do institutions need to take to get there?
The absolute basics towards a perfect data world would include structured data that is easily accessible, discoverable, contextualised, understandable, definitive, and well-designed.
Transparency is good but you can still provide information that people don’t understand, so ultimately the data is still opaque. Putting in the effort to contextualise information is really important. Discoverability is also necessary - even if the information is sensitive, it still needs to be discoverable. No-one wants to hunt through pages of a Google search for answers to a simple question.
I think Parliament could be a better data provider in terms of its users’ needs. There’s a huge wealth of material published by them, but they need to be universally understandable for people with varying degrees of knowledge.
Within those parameters, people need to be able to get definitive information from and about governments, regardless of institutional boundaries. For example, if someone has an interest or issue, the fact that the issue cuts across Ministry of Justice and Ministry of Housing, Communities and Local Government should not be an impediment to them finding what they need. And this is part of the design of data too, to ensure it’s published in such a way that can be easily linked.
Do you think the next generation may struggle to access this data?
You could argue that Parliament has hundreds of years of transparency because it’s published proceedings every day. The question then comes from what form that data is in, where it is stored, and keeping pace with generational change. Generally speaking, the majority of data from state institutions is not structured in a way that can be repurposed across multiple platforms, without diminishing the quality of data itself.
Governments should invest in understanding various audiences, regardless of age, location, generation or background. Understanding that it’s not just one channel, but multiple channels for the same facts, expressed in different ways, to avoid future struggle.
"the majority of data from state institutions is not structured in a way that can be repurposed across multiple platforms, without diminishing the quality of data itself"
At Juro, we’re obsessed with making sure data in legal is structured and machine-readable, so that it can pass between systems and give real insights. Is this a big challenge generally?
The word ‘data’ feels like too broad a term. It’s important to get a greater understanding of various types of data and attributes. High volumes of transcripts in any institution are going to be relatively unstructured, which isn’t a problem depending on what you want to do with them. For example, natural language processing across a high volume of transcripts isn’t impossible. But going to the other end of the spectrum, if you’re in an organisation with a small dataset about organisation structure, it can be a huge problem if that data is represented differently across 5 or 6 different systems.
Finally - what excites you most about the challenge at Citizens advice?
I was told before I joined that Citizens Advice is a fantastic and supportive place to work; people are really passionate about what they do and it’s an excellent combination. The work is really important - it has a positive effect on people’s lives. Having a degree of belief that I can contribute to that in some way excites me.