Start with the hesitation, not the hex code
A customer reaches your product page, scrolls, and leaves. It is tempting to make the main button brighter. The change is small, visible and easy to ship. But it does not tell you whether that customer understood the product, trusted the business or knew what would happen after clicking.
Button colour can matter for visibility and accessibility. The problem is treating it as the default explanation for every weak conversion rate. Before choosing a variation, define the uncertainty you want the experiment to resolve. A useful question is not whether green beats black. It is whether customers can recognise the next step and feel confident taking it.
Research gives the test a reason
Nielsen Norman Group describes how UX research can help teams develop stronger A/B test variations. Experiments compare outcomes, while research helps explain the behaviour behind them. These methods work together rather than competing for a place in the process.
Watch people complete the task. Ask them what they expect from the page. Look at support questions and the points where a journey breaks down. If shoppers repeatedly ask whether a product fits their situation, a clearer explanation might be a stronger test than a new button colour. Do not assume the answer from one session. Look for a pattern worth investigating.
Write a hypothesis someone can challenge
Try this structure: customers hesitate because a specific piece of information is unclear; changing that information should improve a defined behaviour. For example, a product page might explain sizing close to the purchase action because customers struggle to judge fit. The expected outcome could be fewer sizing-related exits, not simply more clicks.
Keep the measurement connected to the goal. More button clicks are not automatically more successful purchases. A clearer checkout entry might improve completion, but it might also shift where people abandon the journey. Decide what success and unwanted side effects look like before the experiment begins.
Small changes still need enough evidence
An experiment needs a suitable design, reliable tracking and enough evidence to support a decision. A small business should not borrow the testing rhythm of a high-traffic platform without considering its own volume. When traffic is limited, direct observation and a focused usability study can be more useful than an underpowered comparison.
Change one meaningful idea at a time when you want to understand its effect. Document what changed and what you learned. If a colour variation wins, ask whether the result is about contrast, hierarchy or something else before turning it into a universal rule for the entire website.
Make the next test worth doing
For an e-commerce team in Melbourne, Auckland, London or the USA, the same principle holds: local audience expectations and commercial context matter more than a fashionable optimisation trick. Review the promise, product information, trust signals and purchase journey together. Then choose the uncertainty that matters most.
At Scott Design Studios, the useful starting point is the customer journey. Bring a page, a business goal and the behaviour you want to understand. That gives a design conversation a better foundation than a list of button colours.