One year ago, we started working with the different Commission policy departments (or 'Directorates‑General') to figure out a new way to structure our navigation and content, organisation-wide.

Our goal was to help people using our site to understand, as Peter Morville puts it, “where they are, what they’ve found, what’s around, and what to expect”.

We’ve never done this before on the scale of the whole organisation. Until now, every department has had its own - sometimes multiple - navigation menus, making it hard for people arriving on the site to get their bearings. Hard enough on a desktop – much harder on a mobile device, as most navigation menus weren’t built to be read on phone and tablets.

From the outset, we’ve based our work on the needs of people who visit our sites (about 17 million people every month), as well as the goals of the organisation. The first level of the architecture is near final and will be used in the Beta (test) version we plan to launch in June.

Challenge of the mega-site

It’s not easy to build an architecture for an organisation as large as ours, with a remit that spans so many policy areas, tasks, topics, countries, departments.

We are what Paul Boag calls a ‘mega-site’ which is “typically owned by a large organisation that encompasses a broad range of services or products and a diverse user base.”

To create more coherence, we've been working closely with the different departments. If you were to drop by one of our workshops, you'd be forgiven for feeling puzzled by our talk of ‘shortlists’, ‘longlists’, ‘twins’ and ‘orphans’.

So this post is an attempt to explain to anyone who hasn’t been closely involved in the process the 10 steps we have taken to building our new information architecture, and why.

1. Collect input on user needs & business goals

In line with our first principle, the new architecture is based on the real needs of people who actually use our websites. We did an extensive ‘intake’ exercise, gathering input from many sources, including from each of the departments, peer organisations like the UN and OECD, and our public-facing phoneline and email query services.

We also scrutinised the Commission’s political priorities, taking key words from work programmes and mission statements. And we looked at our most popular content: what were the main words that people used on search engines and social media to find us?

2. Work the list

This gave us a list of almost 2,000 tasks or reasons people interact with us online. Working intensively with the departments, we whittled it down to just 77.

We ran several workshops with them and spent hours staring at and crunching our way through spreadsheets. The time spent with these colleagues – who are at the coalface of the actual policy areas concerned – gave us a chance to get to know each other and get used to working cross-silo.

3. Ask users

Using the list of 77 tasks, in May 2014 we ran a poll in all 24 EU languages asking people to choose from it the main reasons why they interact with us online.

The response was bigger than we could have hoped – some 107,000 people replied. The priceless insights gave us a task ranking clearly showing us where to prioritise our work. The main tasks? Law, Research & innovation, Funding, Environment, Strategy & priorities, Education & training.

4. Sort results into categories

Taking the top 18 of our tasks, we asked our panel of test volunteers to group these into categories that made sense to them, and to give each one a label.

We then analysed the results, looking for frequent associations between tasks (i.e. tasks that were nearly always grouped together) and common names given to groups. This formed our first hypothesis for a new architecture, based on 15 categories, or ‘classes’, as we call them.

5. Test what goes where

Still working closely with the policy departments, we came up with 30 scenarios - typical tasks people come to our websites to perform. We tested these on 300 users to see where in our proposed classification they would look to perform those tasks.

The first test gave us a success rate of around 60%, but we knew we had to do better...

6. Iterate (test, improve, test again)

Based on the results of the first round of testing, we discussed with the departments on improvements to make. Focusing on those tasks that were still not achieving the required level of recognition in the tests, we tried giving them a more logical home, changing the names of some of the labels, sometimes changing the order of the words.

7. Get agreement

This month we've been busy with our final full session with the departments, where we reviewed the 3rd and final version of Level 1 of the common architecture.

Will it be perfect? No, definitely not. Some categories - like ‘Jobs’ and ‘News’ are proving tricky. But there's no doubt it’s a huge improvement over what we have now, and will genuinely help people get more quickly and easily to where they want to go.

8. Dig deeper – category by category

I mentioned we had 15 classes. For each of these we will be digging deeper to unpack the more granular and specific tasks behind each one.

Under ‘Funding’, for example, people want to find out if they are eligible for funding, whether they have been granted funding, find a partner in another EU country, and so on.

We have started this ‘unpacking’ for the task ‘EU strategy’ - pulling out the long list and reducing it, as described in points 1 and 2 above - and have already had several sessions with the policy departments to look into this. In the coming weeks, we'll be putting out a user poll to help us rank them in order of importance.

9. User experience  and design team take over

The user research team – which has been working on the architecture to this point – will soon hand it over to our UX ('user experience') and design colleagues. It will be their job to turn the 15 categories into a user-friendly navigation which works on different devices, from phones to desktops.

10. It’s so much more than ‘just’ architecture

Making it easier for the user does not necessarily make it easier for the organisation. Nevertheless, the work we’ve done over the last year has helped pave the way for closer collaboration and understanding across departments.

It’s taken the 5-strong team of user researchers some time to come this far. We have enjoyed working so closely with the departments and have got to know each other rather well in the process. All this will help us when we get to the stage of creating content together more collaboratively.

Thanks also to this exercise, we now have a huge body of user research that can help the organisation understand its customers and what people want and expect from us.

We are using the categories to build a governance model which will enable content teams to create content around the top tasks. And the results of our user research and architecture work will also help shape both our search and analytics services and our metadata.


BORDADO, Pedro's picture

I am not a web expert, but I will be pleased to help the EU Comission on financial and economic surrveys.

Add new comment