Large-Scale Site Redesigns Aren’t the Worst Thing in the CRO World

One of the first things we’re taught as CRO professionals is the principle of testing small website design and structural changes in isolation. This ensures that when observing an uplift (or sometimes a drop) in conversion we know exactly what caused it. It makes sense – often after a costly site redesign conversion rates drop suddenly and we’re all left scratching our heads wondering why.

However, one of the most common things we encounter in our job is the dreaded phrase “let’s just change it”. After spending weeks doing comprehensive user research and putting together a laundry list of potential improvements, often stakeholders end up saying “why waste time testing when you’ve shown us that these elements can be improved?”

Normally, at this point, we’d be tempted to explain to stakeholders why this isn’t such a good idea. However, we can often end up in conflict with clients complaining that the process takes too long. This is completely understandable; it is the client’s money funding the process, so ultimately we have to respect their decision.

We might argue the case using the ‘evolutionary site redesign’ principle from Widerfunnel. But, let’s be honest, how many times has anyone actually completed a full site redesign using small iterative CRO changes? It’d likely take years to change a website any significant amount.

 

CRO site redesign

Source

And here’s the crux of my argument:

Sometimes it’s better to just swallow CRO pride and consider that large design changes might not actually be the worst idea in the world.

Why?

Well as Peep Laja explains on Conversion XL, after a lengthy period of testing on a site you’ll come up against your local maxima. This is the point at which tests are starting to see diminishing returns and you’re running out of ideas that are successfully increasing your conversion rate.

Often it’s better to start testing on a site that’s relatively well designed in the first place rather than trying to polish a metaphorical turd.

This example from usertesting.com highlights a redesign that performed extremely well. You’d definitely not be able to do this using small iterative changes:

cro redesign site

Source

We, as an industry, need to understand that we can’t always get exactly what we want. SEO, PPC, UX Design and CRO teams are often competing for their own ‘necessary’ changes to be made and these don’t always mesh well together.

A site redesign is a great opportunity to work with these different teams to change a number of factors on a site and improve it significantly and keep everyone happy.

How? 

So how do we do this in such a way that it doesn’t cause significant problems and end up as an example of what not to do, on a blog post somewhere?

It all comes down to research. The most likely reason considering CRO changes in the first place is due to the strength of your initial research which convinced the decision makers to go ahead and make changes to the site.

Use this to your advantage. Perform user testing, surveys, focus groups and every other quantitative/qualitative research you have available, to put together a definitive guide on what’s wrong with the site and how it can be improved.

In our case, we create a document with as many mini-hypotheses as possible. We create an outline of a redesign and explain every change we’ve planned and why we believe it could improve conversion.

We work with the UX teams and designers using what we’ve learnt and developed a design we’re certain will have a positive impact. We’re still testing it, AB testing it against the old site. This helps ensure that all changes are directed by evidence, rather than blind intuition.

At the end of the day, there’s never a 100% guarantee that drastic website redesigns will be successful. However, doing a redesign using this method is the best way to minimise risk and avoid costly mistakes.

For more information about how Banc Media’s CRO team could help improve your website’s performance, click here, or contact us on [email protected]