The death of Excel? How our community delivered change at Welsh Government
Presenting at the Government Statistical Service (GSS) Conference this year was a real privilege. It was a fantastic opportunity to share the progress we’ve been making in the Welsh Government to modernise our ways of working. I was also able to learn a lot from GSS colleagues about how they are innovating and improving.
For those who missed the session or would like to catch-up, here are the key messages.
Why change was needed
At Welsh Government, our team of 140 statisticians produces over 400 outputs a year, covering everything from health and education to the economy and Welsh language. We’re a small team with a big remit, and there’s little time or flexibility to step away from the day job for transformation projects. We didn’t have extra resources – no new funding, no extra staff – but we did have a community of enthusiastic, skilled people ready to make a difference.
The problem: inconsistency and inefficiency
Our analysts used a mix of tools – Excel, MS Access, SAS, SPSS, SQL, R, Python, and Stata. This led to inconsistent practices between teams, a split in skills, and unnecessary complications for statisticians when they move roles across teams. Manual processes were time-consuming and increased the risk of errors. In addition, cost pressures meant we couldn’t keep paying for a wide range of licensed software.
Our vision: sustainable, efficient, and effective
As a solution, we set a vision. By the end of 2025, we want our analytical infrastructure and ways of working to be sustainable, efficient, and effective. This means:
- using modern analytical tools that fit our needs
- making processes faster, more consistent, and reproducible
- ensuring our work adds value and insight
- prioritising activities that matter most to our users
This meant changing our ways of working to use a small, consistent set of analytical tools (namely SQL, R and Azure DevOps), and to use them to modernise and automate our processes.
How we delivered change: principles and approach
We followed four key principles to ensure this change:
- Community-led: Change was driven by those doing the work, not by a separate transformation team.
- Resource neutral: No extra time, people, or money—just making the most of what we had.
- Harness expertise: We identified and used our internal experts to lead training and support.
- Owned by teams: Teams were empowered to own solutions and shape their own plans.
We designed three workstreams – capability, capacity, and coordination – to deliver change, supported by project management and communications.
Building the foundations
We started by understanding the size and shape of the problem: auditing processes, surveying skills, and developing team plans. Monthly monitoring helped us track progress. We gave teams advice on creating capacity, using analytics, and consulting on outputs. Teams were asked to consider whether their outputs could be reduced or stopped altogether. This helped us focus on what added real value.
The Code of Practice underpinned everything – we focused on user needs and transparency.
Building capability and sharing knowledge
Skill levels in R, SQL and DevOps were varied, so we heavily focused on raising baseline levels of capability. We mapped all available training, categorised it by skill level, and identified internal specialists to deliver new sessions. Recordings were kept for reuse, building our own training library.
Skilled individuals were embedded into teams to work with them to transform processes and replace licensed software, which saved costs and built capability. As our approach to building capability has evolved, we’ve moved to offering flexible drop-in sessions and kick-off meetings with an expert when starting a new project.
Online learning sessions became a real focus, with protected learning time each month. Sharing learning was vital: seeing others start out, asking questions in Teams channels, and supporting each other made a big difference.
Combatting inconsistency was a key driver. We developed new guidance as gaps were identified and created a “Code Catalogue” for reusable packages for common processes, like making charts in our corporate style.
What’s changed so far?
It’s still a work in progress, but the results are encouraging. We’ve delivered around 20 learning sessions, and that learning is being put into practice. Teams report that they are transforming their ways of working, with between 30% and 90% of processes now transformed, depending on the team. Monthly “temperature checks” show most people feel positive about the changes.
Statisticians have really embraced the change. Both experts and those newer to modern tools have shared how much they’ve gained.
“Working alongside another specialist gave me the opportunity to learn new data processing techniques,” said one. “It gave me a chance to share some of my knowledge, to challenge my skills and develop new skills along the way.”
New learners appreciated the mix of online resources and peer support:
“Learning R has improved the way that I work and was made easy with the training resources.”
What have we learned?
My overriding takeaway is that this has been culture change, not just technical change. Supporting people to learn and feel safe is crucial.
You need to create a degree of urgency for the change. Setting deadlines, even arbitrary ones, helps keep momentum going.
As a leader, you have to listen and then act on what you hear. Feedback shaped everything we did, from monthly surveys to open forums.
It won’t surprise you to hear that communication is vital. Be prepared to repeat, repeat, repeat what you say. Share the vision, what’s happening, how you’ve acted on feedback, and do it multiple times and in different ways.
Final thoughts
We delivered real change without extra resources, by harnessing the expertise, enthusiasm, and community we already had. It’s about empowering people, sharing knowledge, and focusing on what matters most.
Thank you to everyone who’s been part of this journey.
If you’d like to know more, please get in touch at KAS.transformation@gov.wales.
Stephanie Howarth, Chief Statistician, Welsh Government