The Problem with Quick Fixes
When something goes wrong, the pressure to act can be immense. We adjust the marketing budget, change the ad creative, and rewrite the website copy all in the same week. Or, in a manufacturing setting, we might alter machine settings, switch raw material
suppliers, and retrain staff simultaneously. The impulse is understandable; it feels proactive. The trouble is, if things improve, we have no idea which specific change was responsible. And if things get worse, we’ve created a tangled mess of new problems, making it nearly impossible to identify the source of the failure. Changing multiple variables at once turns problem-solving into a guessing game, wasting time, resources, and morale.
Step One: A Forensic Analysis of Failure
The most effective problem-solvers resist the urge to immediately change anything. Their first step is to document everything about the current failure. This process, known as Root Cause Analysis (RCA), is about digging deeper than surface-level symptoms to find the underlying issue. It involves gathering data, mapping out the sequence of events, and clearly defining the problem. Think of it like a detective arriving at a crime scene. Their job isn't to immediately chase a suspect, but to meticulously record every detail. This documentation creates a baseline—a stable, understood snapshot of the problem. Without this, any attempted solution is based on assumptions, not evidence. This structured approach helps move beyond blaming individuals and focuses on systemic issues, fostering a culture of learning rather than defensiveness.
Step Two: The Power of One Isolated Change
Once you have a clear, documented understanding of what went wrong, the next step is to form a hypothesis. For example: “We believe changing the email subject line will increase open rates because the current one is not engaging.” This is where the core principle of scientific experimentation comes into play: test only one variable at a time. In an experiment, the factor you change is the independent variable, the outcome you measure is the dependent variable, and everything else is a controlled variable. By keeping all other factors constant, you can confidently attribute any change in the outcome directly to the single variable you altered. If you change the subject line and the call-to-action button simultaneously, you'll never know which one truly influenced the result.
Putting It into Practice
This two-step process is the foundation of A/B testing, a method critical for optimising everything from websites to marketing campaigns. In a classic A/B test, you create two versions of something (like a webpage), but only change one element—say, the headline. Version A is the control (the original), and Version B is the variation. By showing both versions to a similar audience, you can gather clear, unambiguous data on which one performs better. This disciplined method allows for continuous, evidence-based improvement. But the principle extends far beyond digital marketing. It applies to improving a manufacturing process, refining a restaurant recipe, or even figuring out why your houseplants are struggling. By documenting the current state and then methodically testing one change at a time, you move from chaotic guesswork to systematic improvement.
