Dynamic vs Diligent Experiments

Question: Are you running dynamic or diligent experiments?

Dynamic experiments are flexible, resource-savvy tests that focus on achieving simple and practical goals.

Diligent experiments are robust tests that explore complex relationships, providing a scientific inquiry into fixed research goals.

Trained as a Behavioural Scientist, But Not as a Baker

As an economist and mathematician with a master's in behavioural science, I was trained to conduct robust experiments. In my first position as an applied behavioural scientist with Ogilvy Consulting in the UK, I was responsible for designing all the experiments involving behaviour change. We tested the impact of new psychological innovations for clients spanning diverse industries such as finance, health and safety, retail, technology, organization change, and local government.

One might assume that the skills to conduct complex experiments implies proficiency in running simple ones as well. Yet, it wasn't until I started baking sourdough that I realised I hadn't been equipped to conduct dynamic experiments.

Picture: Crumb shot of my delicious sourdough loaf from a 70% hydration, cranberry and lemon zest experiment

A baker’s goal is NOT a scientific inquiry. There is a complex scientific process of interaction between the strength of starter (wild yeast), different types of flour, hydration percentages, and so on.

However when baking, all I care about is how I can make a good loaf with the ingredients that are available.Sometimes I like to add seeds into the loaf for a more nutty flavour. From time to time, I challenge myself to push boundaries by increasing the hydration percentages for a loaf with bigger holes. In many cases, I make multiple changes at the same time. I appreciate this means that my experiment is not going to be scientifically valid. The results will not help me attribute a specific change to the delicious loaf.

A baker's goal is simple and has a practical outcome — crafting a loaf that people love to eat.

"Even if I improve by just 1% with every loaf, the power of compounded learnings goes a long way."

The most humbling aspect of baking bread is that you learn whether your loaf turns out to be a horror ‘frisbee’ (see picture below) or one with a good oven-spring in under 48 hours. Even if I improve by just 1% with every loaf, the power of compounded learnings goes a long way. While the goal with each sourdough loaf may differ, the quick feedback loop ensures that I improve my baking skills very rapidly.

Picture: My horror 'frisbee' sourdough loaf

When the sourdough is perfect, the whole loaf is often consumed in one sitting. If my bread isn't up to par, I notice how much of the loaf ends up in the food compost pile. This quality of rich behavioural data is a scientist’s dream. In short, you will never get to improve the quality of your product unless people get the opportunity to taste it.

Reliable is Not Always Useful

As a behavioural science consultant, I conducted numerous scientific experiments, ranging in duration from 2 weeks to 2 years. In some instances, after the robust tests concluded, the entire client team underwent significant changes, and the new management showed no interest in the original research goals. In other cases, technological changes prevented them from adopting our scientifically valid results into the new product offering.

In hindsight, most experiments I designed were focused on scientific reliability, which stems from my training as a good researcher. Now, as a sourdough baker, I've discovered that prioritising the speed and flexibility of experiments is often more valuable than being diligent.

Learning the art of baking has enhanced my skills in dynamic experimentation, not only as a behavioural science consultant, but also across all my work as a small business owner, science writer and breathwork facilitator.

A baker’s recipe for designing experiments

  1. Set your goals: Before you start an experiment, always ask yourself: “what does good look like?”
  2. Outcome is key: Run a dynamic experiment, changing multiple variable, if you believe it can improve outcomes for the people you serve.
  3. Feedback loops: Focus on the speed of learnings from your experiment so that you can get 1% better, a little quicker.

Keep learning insights from behavioural science

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Learn something new every month. We deliver behavioural insights straight to your inbox.