Improving Travis Perkins navigation structure for improved conversions

by David Mannheim

posted on: March 2nd 2018


Many ecommerce websites suffer from the same problem: too much categorisation.

An exercise such as card sorting would work wonders to revisit the categorisation of your website, however, after several sessions of moderated user testing and screen recording analysis using FullStory we identified how users wanted to navigate, not just how users were navigating.

Looking at how users ‘were’ navigating is a bit folly; because they are doing something within the realms of what is possible for them. When speaking to users within moderated user testing sessions, it was quickly noted that the 250 categories were overwhelming for users for a myriad of reasons:

  • Merging category names together eg Category X + Category Y + Category Z all as ‘one’ category
  • Bundles of categories within bundles of categories as a result of the above
  • Grandchild category structure (categories within categories within categories)
  • Tenuously categorised categories eg. “Timber” which could be under Timber or Joinery

This added friction and, this, the length of time it took for users to reach their desired category.

But how did users want to navigate? Well, we didn’t know without trying it – that’s the purpose of experimentation. After all, users don’t know what they don’t know. With an over-categorisation of products, we can look elsewhere for inspiration. For us, this was offline. Specifically, Argos and the use of the indexation within their catalogues. Why don’t Argos use this indexation online, we thought? With so many products and so many different categories, providing a simple A to Z of categories would surely be easier and quicker for the user?


With that hypothesis we amended the navigation to include an A to Z listing.

The results of this were that we saw:

  • an increase in conversion rate of 26% of users that used the navigation at a 95% statistically significant result over a period of 4.5 weeks
  • Interestingly, 40% of users utilised the A to Z listing and these users converted 7% higher than those users that clicked on the incumbent menu.

We analysed the number of times users hovered over the menu link and then the A to Z link (suggesting they couldn’t find what they wanted within the menu), clicked the A to Z link thereafter (suggesting the A to Z listing suited their needs), the impact on both these metrics and overall metrics against the search bar to name but a few.


When reviewing session recordings, however, we identified that of those users using the A to Z listing, users were clicking on the anchor links at the top to get to their desired letter and were unable to find their product. Continuing their journey and relaying this information back, the category was there just named differently. Take “decking” – some users call it “Timber Decking” opposed to just “Decking” (so, “T” instead of “D”)

We then added many more categories in the A to Z listing to account for any form of mis-categorisation.

This again saw an uplift across all relevant metrics by a further 5% of users using the A to Z listing.


We still, however, saw the same pattern within the listing; users unable to find their product category by first letter alone; again through analysing screen recordings (there is a more specific case study on screen recording for optimisation here).

We then added in a search bar to overcome this challenge which you can see below. This addition resulted in a further 3% increase in conversion rate – interestingly, the feature was used sparsely and an option for future iterations.

Which we changed to:


Over-categorisation is an issue with ecommerce websites offering a broad range of products.

Whilst card sorting can go some way to addressing this issue, the display of those categories is equally, if not more important.

Think outside the box. It’s not about doing what others ‘haven’t done before’ but about researching your customer needs and serving solutions that match those needs.

The final result? This experiment has, so far, undergone 4 iterations and to date generated over £380,000 worth of incremental revenue for Travis Perkins in the experimentation period alone. … And we still have future iterations worthy of running on the navigation alone.

User Conversion went to town on our navigation structure. Instead of embarking on a lengthy card-sorting process that would have taken us months to re-categorise and implement, we chose to redesign the navigation; how users interacted with it. The return that we saw from this experiment, alone, was phenomenal.
Adrian Hobson Online Trading Manager
avatar for author

David Mannheim

David is an experienced conversion optimiser and has worked across a series of core optimisation disciplines including web analytics, user experience and AB & MVT testing.

Improving Travis Perkins navigation structure for …

by David Mannheim Time to read: <1 min