One small change had a million-pound impact: “9 out of 10 people in your area have already paid their taxes.” With this phrase in its payment letters, the UK tax authority (HMRC) managed to get thousands of citizens to settle their tax debts. The result was £1.2 million in additional revenue above expected collections in just one month.

No new laws, no higher fines: only the communication was adjusted using behavioural science.

At our latest Gobe Ventures Breakfast, we asked Marta Garnelo, Principal Advisor at The Behavioural Insights Team (BIT), how this discipline can be incorporated into the field of public administration. Marta has more than a decade of experience advising governments and international organisations on how to transform their policies. She has led BIT’s offices in New York and Latin America, in addition to collaborating with institutions such as the World Bank and J-PAL at MIT.

Her proposal points to a shift from intuition to data: designing policies based on how people actually act and make decisions. By replacing assumptions with evidence, public administration can simplify citizens’ lives and improve the quality of public management.

This vision is built around three strong ideas. First, complementarity: behavioural science does not replace traditional tools, but makes them more effective. Second, evidence as a guide: each intervention is based on data and tested through a method that determines whether it actually generates savings or improves a service. Third, operability: frameworks such as TEST (Target, Explore, Solution, Trial) make it possible to move from a problem to a clear roadmap.

Below, we summarise some of the key ideas for applying this discipline and generating impact from the very first interventions.

1. Understanding How We Decide

This discipline acts as a bridge between psychology, economics, and public policy design. Its operational basis lies in Daniel Kahneman’s dual-process theory, which explains how we operate through two mental systems:

• System 1: intuitive, fast, and automatic (such as catching something that falls or knowing that 2 x 2 = 4).
• System 2: slow and reflective; it requires effort and handles more complex processes (such as analysing regulations or multiplying three-digit numbers).
  

Public administration often designs services on the assumption that citizens operate through System 2. However, in the real world, people rely heavily on System 1 due to inertia. Understanding this reality makes it possible to work on choice architecture: the deliberate design of context so that a given behaviour becomes the easiest or most appealing option.

Actions as simple as making digital payment the most convenient option on a screen, or using messages based on social norms — such as reminding people that the vast majority of their neighbours have already met their tax obligations — help ensure that the desired action becomes, quite simply, the most natural one.

Behavioural science became institutionally established worldwide following the publication of Nudge (2008) and the creation of BIT in 2010. Today, hundreds of government units around the world apply these techniques to improve everything from tax collection to public health.

‍

2. The Test Methodology: An Operational Framework for Innovation

BIT’s TEST framework offers a roadmap for structuring an evidence-based intervention:

T — Target. Define the target behaviour. In other words, specify who should do what, when, and where. It is more effective to focus on specific behaviours, such as making an online payment, than on trying to change broader habits.

• Example: rather than aiming for “people to use the online portal more,” aim for “users over 65 to complete the online payment form.”
  

E — Explore. Understand the context and identify where the process breaks down or where the bottleneck lies. This involves analysing quantitative and qualitative data to determine, for example, on which screen users abandon a process.

• Example: in AI adoption, analysis may reveal that the barrier is not a lack of training (capability), but fear of being replaced (motivation).
  

S — Solution. ‍The EAST principles (Easy, Attractive, Social, Timely) are an international standard for designing interventions:

• Easy: reduce friction and simplify action. A newly simplified payment card tripled online payments in Singapore.
• Attractive: capture attention in an intelligent way. A chatbot in the province of Chaco, Argentina, tripled booster vaccination uptake.
• Social: use group influence and social norms. This is the logic behind the phrase “9 out of 10 already do it,” used to reduce tax delinquency in the UK.
• Timely: intervene at the exact moment a decision is made. Sending alerts at the right time reduced referrals to overcrowded hospitals by 38%.
  

T — Trial. Evaluate and learn. Through rigorous evaluation, such as randomised trials or A/B testing, a group receiving the intervention is compared with another following the usual process in order to identify the real impact before deciding whether to scale it.
‍

3. Where To Start: Some “Quick Win” Guidelines

For those looking to take a first step in applying behavioural science to policy design, it is important to understand the variables that affect the cost and duration of this kind of project. These include the type of behaviour one is trying to change, as well as the complexity of the evaluation process.

Marta Garnelo shared three broad categories for setting expectations around a project:

• Low Complexity (~1 month): interventions in digital communication, such as testing messages on a website. This is the ideal entry point because it makes use of technical tools that already exist.
• Medium Complexity (6-18 months): redesigning a complete process. The duration depends on user volume; if a procedure has low traffic, it will take longer to gather representative data.
• High Complexity (+18 months): large-scale evaluations testing multiple solutions simultaneously. These are long-term bets aimed at structural transformation.
  

How should you choose a good candidate for a first project? Some useful guidelines include:

1. Specific actions. It is much easier to influence a single action, such as choosing a payment method, than to try to change a habit, such as recycling, or solve a major challenge, such as obesity.
2. Clear return (ROI). Prioritise projects that generate obvious savings. A classic example is encouraging the shift from in-person to online procedures, which reduces staff workload.
3. Existing data. “The most expensive part of an evaluation is obtaining quality data.” One recommendation is to make use of the information that public administrations already collect on a daily basis.
4. Autonomy. Coordination across different administrations is valuable, but often slows things down. A good place to start is where the team has direct control and does not depend heavily on other departments.

From Intuition to Evidence

Behavioural science does not replace traditional tools; rather, it acts as a design layer that makes those tools more effective by connecting them to how people actually behave. It is no longer a matter of guessing what citizens need, but of identifying the barriers that prevent them from accessing a service and the interventions needed to overcome those barriers.

Public innovation therefore moves beyond the realm of “good intentions” and begins to rely on diagnosis and evidence. This transition requires methodological rigour. As Marta Garnelo pointed out during the session: “If it weren’t for rigorous evaluation, no one would have believed that intelligent design could have such an impact.”

Adjusting the design of public policies based on the lessons of behavioural science has enormous potential to improve the relationship between public administration and society, which is one of the key goals of digital transformation.

At the end of the day, the purpose of innovation is to build a more humane and pragmatic state, where efficiency is synonymous with usefulness. A state capable of designing processes in which behaviour that benefits everyone is simply the easiest path to follow.

‍