Behavioural science, the study of human behaviour, encompasses neuroscience, cognitive science, sociological, and psychological disciplines in order to understand and predict how humans will act in given situations. This involves observing behaviour, identifying patterns in it, and backing up insight into the reason behind those patterns using data and rigorous control. Sound familiar? Well, it should!
Behavioural science and CRO
What did we ever do before Google Analytics? Pre-2005, launching a website, or publishing a new landing page was a kind of hit and hope activity, but good ol’ GA fixed it so that we could start gathering user behaviour data. We got to know where visitors came from, whether they visited certain pages, whether they completed a form – we could track and measure how people were interacting with our websites… FOR FREE! No longer would webmasters need to guess what visitors to their site wanted to see, and no longer would they just throw stuff at the wall to see what sticks. Perfect website here we come – except that shifting from making decisions based on a hunch, or on “best practice”, to making decisions based on research can be tricky. Time for an example!
Marks & Spencer’s site redesign
In 2014, M&S launched a £150 million website redesign. The project took two years and while ExperienceUX described it as “a positive redesign” (they had a whole 21 reasons why they liked it!), customers did not feel the same. The launch of the new site resulted in an 8% decrease in online sales but a huge increase in frustrated customers.
Despite the efforts to improve navigation, search functionality, and filtering of products, the new site brought about several massive usability issues that just couldn’t be ignored:
- They scrapped their user database in the redesign. That’s a lot of people having to re-register all their personal information like payment details and delivery preferences, and that’s going to cause some people to abandon their cart.
- It was buggy as hell from the off. The site crashed on launch, stock numbers were inaccurate, customers struggled to complete their purchases, and if they did manage to checkout, their packages were being delivered to addresses they hadn’t selected as their delivery address. Imagine the impact that would have on people’s trust in the brand!
- It was a complete overhaul. For regular customers who had been using the old site for years, the new UI was too much of a change. They’d learnt how to use the old site, but would have to forget all that and re-learn it.
The wise way to redesign a site is by using behavioural science
At a whopping £150 million, you’d expect that M&S’s team undertook some serious market research, conducted some serious usability testing, and engaged in some serious behavioural science; but with a two year time scale, did they account for changes in trends and user preferences? Probably not, as the outcome would suggest.
Take Amazon, the champion of usability testing: They make changes and tweaks to their pages all the time, but they know their customers inside out, and rigorously test any changes on a sample of visitors before launching them site-wide, allowing them to reduce the risk of any changes being a disaster of M&S proportions. Instead of a group of experts coming up with how they think the site should look, a group of experts looks at user data, applies the principles of behavioural science, and builds a testable hypothesis.
A defined CRO process
Redesigning a site should be a commitment to learning what your customers really want, and that rarely happens overnight. Think of conversion rate optimisation as a way for your website to evolve over time:
- Identify problem areas. Whether it’s Google Analytics, or another service, gathering user data is essential in discovering where your user journey bottlenecks.
- Test your theories. If you’ve got ideas for improvements, test them. Colour, layout, copy; a simple a/b test will help you understand how these elements impact user behaviour.
- Go again. Each test, whether your idea “wins” or not, is an opportunity to learn – it’s basically another source of user data. Apply your learnings to more idea development, and more testing.
Use what works and reject what doesn’t.
Also published on Medium.