Testing & Learning in the Physical World

In his book The Behaviour Business, our BVA Nudge Consulting colleague Richard Chataway highlights the importance of “test tube behaviors.” By this, he means that companies need to adopt a test-and-learn culture, in which they are continually introducing new ideas and learning from the actual behavior of their customers. To illustrate this point, he shares examples from Google, Amazon and Netflix, illustrating how these leading companies of 21st Century are applying behavioral science – and an ongoing commitment to experimentation, testing and data-driven decisions – to optimize services and encourage/influence specific customer behaviors.

Adopting a test-and-learn mentality and a data-driven culture is also critical from a more practical standpoint: Companies need to quickly gauge the impact of new ideas and interventions, in order to project their return on investment (ROI) and justify investments in rollouts. Without this process in place, many promising ideas and “nudges” will never be introduced.

Yet examples from the tech giants can only take us so far. That’s because these organizations have built-in advantages, as they are fully digital businesses with access to millions of customers. Thus, it is relatively easy for them to run continual A/B tests (on different home pages, offers and wording) and generate data in real-time.

But what about businesses in other sectors, such as Consumer Goods, Transport, Hospitality or Financial Services, which are running their businesses largely in physical environments?

  • How can they test and learn in an agile way, when they don’t have immediate access to millions of people and digital transactions?
  • How can they quickly evaluate new interventions to gauge their potential impact, improve/refine their execution – and justify their introduction?

Related: Introducing test tube behaviours – the mindset | Part 1

Of course, there’s not a single answer, as businesses that are both physical and digital require a toolkit of test and learn approaches. But from our experience, we can offer several guiding principles:

Consider the nature of the intervention itself

Digital interventions (such as texts, apps, messaging, etc.) are perhaps the simplest to consider, as they can often be A/B tested, either “live” or through controlled online studies.

Physical interventions, which will appear in stores, homes or other environments, can be more challenging, yet there are several approaches to consider:

  • Simulated studies (i.e. virtual shopping) can often be used to pre-test many interventions that will appear in a retail environment (involving signage/messaging, packaging, etc.).
    • For example, we recently used our ShopperLab (i.e. a simulated grocery store) to pre-test retail “nudges” for a Consumer Goods client.
  • Piloting (in a limited number of stores, branches or homes) requires more time (and typically cost), but provides the benefit of gathering actual customer data
    • For example, we are currently piloting numerous interventions on actual trains for a railway client.

The commonality across these approaches is that we are simulating or introducing a new situation – and documenting its impact on behavior. We are watching and recording what people actually do, rather than relying on what they say or claim.

Calibrate the rigor to the risk involved

In the academic world, where most Behavioral Science learning is rooted, the primary objective is to learn and determine causality. Therefore, its critical to isolate variables and document statistical significance, via randomized control trials (RCTs). But in the business world, where the goal is typically to guide a decision and/or to justify an investment – and cost/timing are always considerations – this level of rigor isn’t always necessary or advisable. In fact, the beauty of many nudges is that they are inexpensive to develop and execute. They don’t represent enormous investments nor risks – and thus, it doesn’t make sense to spend time and money testing them, nor to isolate the impact of each individual intervention. In these low-risk scenarios, we may recommend introducing combinations of nudges, launching them (via pilots or A/B tests) – and learning/adjusting from this real-world experience/feedback.

For example, through a series of “nudges” and A/B tests, we were able to increase the conversion rate at a non-profit web site from 2% to 24%.

Build “test-and-learn” thinking into the ideation and screening process

While businesses can (and should) be pragmatic in their approach, they should also be systematic in their processes. This means incorporating testing plans into their behavioral science initiatives from the start. This is not only a question of project planning, timing and budgeting, but also of how new ideas/interventions are developed and screened.

When creating nudges, there’s a need to outline how they will be tested, piloted and measured. In fact, this next step should be a criterion in evaluating a wide range of ideas – and selecting those in which to prioritize and invest further resources. After all, it makes no sense to invest time and energy in developing interventions, without a plan for documenting their potential impact – and justifying their introduction.

Related: Introducing test tube behaviours – the methods | Part 2

“Test Tube Behaviors” Make Behavioral Science Actionable

When most organizations think of behavioral science, they focus on the excitement of identifying new interventions, rooted in human heuristics. But adopting “test tube behaviors” – a commitment to experimentation and continually learning from customer behavior – is an equally important part of the equation. And while not all companies can be Google, Amazon or Netflix, they can adopt this mentality and apply it pragmatically, adopted to their own reality. In doing so, they will help ensure that the insights of behavioral science translate into successful interventions – and don’t “die on the vine” due to a lack of evidence or proof to support their introduction.

This article was written by Scott Young (scott.young@bva-group.com), Senior Vice President of the BVA Group, and Ted Utoft (Ted.Utoft@bvanudgeconsulting.com), Vice President of the BVA Nudge Consulting globally.

Recent Headlines from the BVA Nudge Unit


The Behavioural Science Behind Gift-Giving

Michelle Novellie

BVA Nudge Consulting

We’re often reminded that festive holidays are about more than just gifts – but we don’t always appreciate just how […]


The Reluctant Zero-Sum Consumer

Koen Smets

Senior Advisor, BVA Nudge Consulting

What distinguishes humans from the rest of the animal kingdom? Among many possible answers to this question, there is one […]


Impossible Foods' Success: A Behavioural Science Perspective

Written by Richard Bordenave with the support of Ryan Xie

People love meat. Yet CEO Patrick Brown aims to end all meat consumption by 2035 through Impossible Foods, his company […]