Understanding behaviour: designing conditions where test and learn thrives
For many organisations attempting to implement test and learn, the biggest challenge they face isn’t one of method or organisational intent. Most people understand why small iterations are better than big launches, and why testing with users beats guessing. Often, the problem isn’t awareness – it’s behaviour.
Whether it’s asking civil servants to adopt new ways of working, customers to use unfamiliar digital services, or leaders to embrace uncertainty, introducing test and learn involves asking people to change how they behave.
When transformation efforts zero in on processes, tools and training, they overlook a crucial truth: humans don't behave the way traditional management theory assumes. Our actions are shaped by our unconscious biases, habits and social cues.
As behavioural science has shown for decades, we tend to stick with what's familiar, copy those around us and favour short-term wins over long-term gains. It's no wonder so many digital strategies look so shiny on paper but falter in reality – they're built around how we think people ought to act, not how they actually behave.
Working with human psychology, not against it
Test and learn is a powerful way to embed lasting change. But it also clashes with some deep behavioural and cultural biases. Most of us feel safer with plans, control and clear predictability.
The key isn’t to fight these instincts but to design with them. By understanding how people actually behave, we can shape environments where better behaviours happen by default.
At Public Digital, we've seen this first-hand across dozens of transformation efforts. The organisations that succeed don't just mandate new processes – they create the conditions where test and learn behaviours emerge naturally. They make the right things easy and the wrong things hard.
Many of the approaches that work best in transformation align closely with behavioural science, even if they’re not always labelled that way.
Here are five that make a consistent difference:
1. Celebrating small wins to overcome fear of uncertainty
Most of us instinctively avoid uncertainty. When faced with complex challenges, we typically either wait until we feel "ready" (which rarely happens) or create elaborate plans trying to predict every detail (which almost never works).
Test and learn works better when we break big changes into small, low-risk experiments. By creating early wins and showing visible progress, we tap into how confidence actually grows - through a series of small successes rather than one perfect leap.
When Universal Credit was reset in 2013, the team started with just 20 users in Sutton before gradually expanding. This wasn't just good delivery practice - it aligned with how people actually process change and build confidence.
Similarly, Netflix didn't try to disrupt the entertainment industry overnight. They began with DVDs in the post, testing and learning as they went.
In each case, these organisations avoided the trap of trying to launch perfect solutions all at once. Instead, they made it safe to test ideas early and build on what actually worked.
2. Making the invisible visible
You can't learn and respond effectively if you can't see the results of your tests. Our brains pay most attention to what's right in front of us, not abstract concepts or distant outcomes.
During the COVID-19 pandemic, the UK's testing programme used prominent displays showing tests processed and turnaround times. This simple change helped teams focus on test results reaching people in time to be useful (an outcome) rather than just raw testing capacity (an output).
Don't just talk about user needs and outcomes - make them impossible to ignore. Put quotes from your users in team spaces. Share real stories in meetings. Celebrate teams who improve real outcomes, not just those who hit delivery milestones. When we can see immediate feedback from users and the impact of our changes, we naturally adjust to make things better.
3. Moving from blame to learning
For test and learn to work, people need to feel safe taking risks. But in most organisations, there's a powerful force blocking this: fear of being blamed when things go wrong.
Traditional ways of holding people accountable often punish perceived failure and reward following processes. This creates an environment where avoiding risk matters more than learning, which kills the very experimentation that test and learn requires.
The fix isn't complicated, but it requires intentional change. When things go wrong (as they inevitably will), focus on what was learned rather than who is at fault. Create team charters that explicitly prioritise learning over blame. Have leaders visibly celebrate valuable lessons, not just successes.
We've seen organisations dramatically increase experimentation by replacing audits focusing on failure with more positive learning reviews. This simple shift creates the conditions where teams feel safe to take the calculated risks that test and learn demands.
4. Overcoming our tendency to hide
Test and learn thrives on transparency and shared learning. But let's be honest - most of us feel more secure hiding our work until it feels finished. This tendency slows down learning and prevents others from contributing valuable insights early on.
When the Government Digital Service made all its code repositories public by default and ran weekly show-and-tells where teams shared work in progress, they weren't just following good practice - they were reshaping social norms.
The key is making sharing easier than hiding. Use tools that make working in the open the default. Have leaders share their messy first drafts. Create team rituals that make early sharing normal, not scary.
5. Funding teams to create space for learning
We consistently underestimate how long things will take, something behavioural scientists call the planning fallacy.
Project-based funding makes this worse by creating artificial deadlines that force teams to deliver predetermined outputs regardless of what they're learning. By funding stable teams rather than time-limited projects, we create the space for genuine testing and learning. Teams can pursue outcomes rather than outputs, adjust based on what they discover, and build on their knowledge rather than disbanding just as they're gaining momentum.
This approach recognises that digital services are never really "finished". More importantly, it creates space for teams to test approaches, learn from users, and continuously improve rather than rushing to deliver predetermined features before the money runs out.
Making test and learn inevitable
Test and learn might sound like common sense - of course you'd try things, see what works and adjust. But it represents a profound shift in how organisations approach uncertainty.
If there’s one thing I’ve learnt from applying behavioural science, it’s that real transformation isn’t about forcing change through sheer will. Nor is it about finding perfect solutions or quick fixes. Instead, it's about creating environments where testing and learning feels natural and inevitable.
Written by
Camilla Devereux
Principal Consultant