Strategic optimisation: It’s about behaviour, not buttons

Most decision-makers drift away when something slightly difficult to understand. That happens with conversion optimisation, and this attitude comes with a cost. “That’s for specialist, geeks and engineers”, I feel them thinking. And they are partially right: there are just a few hours a day for a decision-maker define where everybody is going. Who has time for tactical work (for example, making one single project better) when one can be dedicated to strategic work, and make all new projects better.

But that’s the catch. From my experience with CRO (conversion rate optimisation), the job is always tactical, but the work — and what you learn with it — is always strategical.

And that’s also the fascinating side of it. First, you get to see what you are missing out by not experimenting, constantly, frequently and even better, continuously. It’s short-sighted to perceive the activity of experimentation as merely changing the colours of buttons. Sure, you can change colours of buttons and get a tiny increase in clicks. But once you increase — even a little — the impact of these experiments, you start to have as takeaways the understanding of consumer behaviour. And that includes principles, motivations, openness and elasticity.

Take, for instance, one experiment we have conducted in a daily-goods, grocery store type of client. When adding something to their cart, the users were simply informed of their product-to-cart addition with a red notification on the top left corner of the page. They were then staring at the product they have already bought. Having time to think if they really need it? Calculating how much they could be saving if not purchasing at all? Uninspired of what to do next? Yes, uninspired of what to do next.

We decided then to add a pop-up window informing what they have added, and suggesting related products, insurance and other add-ons. Basic stuff. At that point, Amazon was already taking users to a totally new page after purchase, totally dedicated to based on what you bought offers. But we decided on the modal, pop-up window.

What happens next? It flops. Users were hitting the Go to checkout button right away, making it easy to check out. Was that an indication that the site sucked so much that users wanted to get out as fast as possible?

We then changed the layout of it. We made “Continue shopping” a dominant button, and “Go to check out” as faded, small text under the button. And it worked. It worked so much that the modal window became best practice in our team, offered to a number of clients in other eCommerce categories.

We had a significant increase in conversion rates, maintaining the average order value. Now that’s great optimisation. We ended up increasing the revenue of the client in about 100K per year. All that done with a few weekly hours of experimentation, and the willingness to let us do the work, and be patient, because false indicators are common if you don’t let the experiments run for enough time.


On upcoming posts, I will talk about the major pain points, barriers and difficulties in running experiments, as well as the future of it with AI.