John Lewis :
Improving sales through Information Architecture
Overall
Information architecture is a profound art that, through the correct categorisation and organisation of hierarchy beyond navigation, can transform an interface from confusing to easy to use without any visual design. In the case of John Lewis, with next to no technical changes, we increased sales in many areas by over 10%.
Problem
This project was a mixture of business strategy and issues raised by analytics and merchandising teams. The plan was to improve the sale of fashion items online, and the analytics also highlighted problems with people finding electrical items; existing labels were Technology and Electric Appliances, with clearly confused users.
We also needed more engineering capability to make the changes, and there were other design projects, which meant the changes would have to be made through the site configuration.
The IA changes would be 2-3 levels deep (5 levels deep in the garden section) and provide flexibility for adding and removing individuals. We also wanted to increase brand visibility. The only area we should have touched was the gifts section, which was working well and worth a separate project. The faceted search was open to improvement but needed to be more central to the project.
We used the realworld store experience to highlight the issues with the virtual store
John Lewis had very good curated sales figures to work from
Our analytics team also had the site well tagged which we used to explore the journeys customered used
Engaging
The key people I worked with were the merchandising teams, who spent their days putting items on our virtual shelves. Throughout the project, I worked continually with them as I advanced the new overall site structure. I used simple visualisations to show how the hierarchy worked and the information we got from tree testing with end users. The analytics team was key to helping me uncover behaviour among the tens of thousands of visitors daily and measuring success.
Discovery
I rolled up my sleeves and mapped out the whole site structure for John Lewis online, stopping before individual items and the structure of the facets—this would be the primary navigation only. I created a visual version of the existing map with different degrees of detail, but the most helpful element was a spreadsheet with all the categories and issues collected.
I also visited several John Lewis stores and talked to staff and customers about the issues with finding specific items in a physical store.
I could not solve the IA for each item, so I used three main categories: heroes—items that made much profit and sold very well; workhorses—items that were expected, dependable, and kept the lights on; and, lastly, problem children. These items could either fit in multiple places or did not naturally fit in any one location in the store, such as Breadmakers and Dress-Making models.
I worked with external agency to provide rounds of tree testing with user thoughts included
Tree testing proved to be the most direct and effective way to find the best way to
We used archetypal products to stress test the structure of the online store
Later we would also explore how content worked with the overall structure
Instead of personas we used user types with real example scenarios
Instead of personas we used user types with real example scenarios
Delivery
As I developed a new categorisation method, I worked with each merchandising team. Then, I worked through a series of usability testing rounds using the example items to drive tree testing in a lab environment. The visual design was about ensuring the items fit in the navigation, and we had one technical ticket to remove the ‘men’ and women prefixes from sub-levels that reduced how scannable the navigation was. We also added brands to the drop-down menus. The final proposed structure is explained in a PowerPoint deck, and the details are provided in a spreadsheet designed to allow the merchandising time to make the changes quickly. We then did A/B testing, starting with low numbers with analytics and keeping an eye on it to see if anything broke.
The heart of the project was an spreadsheet that contained all the structure, notes and changes.
Along side this was a set of overview slides to communicate to the wider time
We showed the overview of the changes
And for those interested there was a visual version of the details
Later the IA project fed into the first mobile experience for John Lewis I designed
My knowledge of the site navigation allowed me to customise the mobile navigation to find items quickly
Results
We set out to increase sales by improving the navigation’s findability, and we delivered on every significant aim.
Fashion items were much easier to find with one or two clicks removed from all journeys.
Users no longer flipped between Technology and Electrical Appliances by merging them into Electricals and adding subcategories.
Game console sales increased 100% at one point (it was one of the most hidden areas)
Beauty, a section most visible in the store, now appeared at the top level, bringing sales closer to those in the physical stores.
Overall sales improved, brands were easier to access, and more people made it further down the funnel.
Learn
Card sorting has limited uses in IA. Although controversial, we used it for an internal workshop, tried it with users, and did not gain any solid insights.
Tree testing is more effective than building prototypes for site navigation.
The physical world and virtual world are not so different, and one informs the other.
Analytics needs to be set up to follow user browsing, not just focus on the funnel.
People enjoy working with people who have an overview and empower them. The buyers have years of experience but trust me to improve their world.
Design Portfolio V5.9
Created using Framer