A plea for the lost practice of information architecture

Vicky Teinaki -

Can you imagine what it looks like to build something purely incrementally without an overall plan? While a lot of older cities are like this, more recently there is a house that shows the results of this approach. 

The Winchester Mystery House in San Jose California is described as the world’s largest home renovation. In 1886, the new owner (and newly widowed) Sarah Winchester inherited an 8 bedroom house and started renovations. By the time of her death in 1922, the house had mushroomed into a total of 160 rooms, 47 stairways and fireplaces, 6 kitchens and more. While there is a popular myth that the constant building was based on superstition - which inspired a 2018 horror film starring Helen Mirren - other accounts suggest instead that she was a strong-willed and particular homeowner, choosing both to eschew architects and redesign the house room-by-room, and to abandon work or have it rebuilt if it did not meet her standards. (In fact Nathan Shedroff pointed out to me that this attraction was perhaps not the house she built but in fact an attraction created after an earthquake… but I digress).

No matter the reason, the house is a memorable attraction but hardly a habitable space. It has a maze of rooms and staircases that range from the confusing (dead ends, rooms with multiple doors next to each other) to the dangerous (a door opening into a two-storey drop into a sink). 

Example from Information Architecture: Blueprints for the Web, Christina Wodtke, 2002

I only found out about the Winchester Mystery House last week, but it captured a phenomenon that I’ve become increasingly aware of in recent years—haphazard information architecture at the least setting off my spidey-senses on consistency and at worst blocking users from getting the thing that they need.

Times have changed

I’ve been designing websites since the late 2000s - long enough to remember having to optimise for Internet Explorer 6. Those early years were in agencies doing a mixture of brochureware sites, e-commerce (some big clothing sites) and important transactional services (a few different types of pensions as well as some big pharma sites). 

When brought on for a project, some of the early things I did as part of my job was:

  • Do a content audit and create a content inventory (using site crawler tools like Screaming Frog SEO and maybe also look at analytics)
  • Create initial site map and navigation (header, footer and so on)
  • If possible also do card sorting and tree tests as part of some of this process (not always a given in agency work)

All of this I’d call information architecture work. 

However, when I moved into in-house agile delivery work in 2015. I found that much of this practice seemed to disappear. Pretty much every existing service I have worked on since - including one service with nearly 200 screens and several with fairly important account functionality - came with no site map and so I had to reverse engineer the service to make one. No one had prioritised it as anything important. I usually did a lot of side-of-desk effort for months compiling an inventory using test environments and Google Analytics page reports. 

While this could have been justified as just ‘working software over documentation’ (as per the Agile manifesto) I sometimes found that doing the sitemap showed bizarre setups that—possibly due to project swerves, slow building on existing infrastructures or both—just didn’t even look usable. One example particularly sticks with me, where the landing page links and navigation with and without a saved form were so different that they might as well have been different sites. I wasn’t surprised that there were a lot of phone calls about people going on to the site and not being able to find what they needed .

I am also not the only person who has noticed a lot of UX work that is done without seeming any knowledge of information architecture and its related skills of structuring information (hi Andy and Dean). While my inhouse agile experience has solely been in the public sector, I also think that this equally affects the private sector and public sector (though in the UK at least it may have manifested in different ways). 

How I think this happened

Having watched this happen over the last 15-20 years, this is my take on the series of events that have led to the erasure of information architecture.

Late 2000s: information architects becoming user experience designers

In the early days of the internet, everyone was an information architect mostly. There was some excellent work going on at places like the BBC Radio Labs - thanks Paul Rissen for the link). However, at the IA Summit 2009, talk Jesse James Garrett's plenary heralded a change when he said ‘it’s all UX’. This means that if you talk to a lot of UX or design strategy people these days who have been in the industry for a while, they may well have had a job title change in the 00s. Some of this was good—it also meant more of a focus on user research—but it does mean that a role that was based in library science and classification disappeared.

Early 2010s: agile delivery and GOV.UK

While Agile and related methods such as XP and Scrum have been around since the 1990s, what really saw this move into the mainstream to the point of affecting UX I think was the popularisation of product management, lean startup, and through this Lean UX. 

However, while the elements of delivering in increments caught a lot of attention, what was often missed was the more subtle ‘outcomes over outputs’. There was also less discussion of techniques such as the ‘walking skeleton’ of storymapping, where some general thinking is needed but then mapping on how to do this in a considered and extendable way.

I think that a lot of this is captured in the often shared diagram of how to build in lean startup style.

The common product management diagram - but how does one get from the skateboard to the bike to the motorbike… 

However, what no one talks about is exactly how someone gets between the stages, and the related rework and more detailed planning. Even techniques such as Wardley mapping assume that elements can be split apart and bought outright rather than reach a point of still being built but requiring more architectural underpinning. Without this reset and rework, everything can end up as another Winchester Mystery House. (I’m not the only person to write about this: Bram Wessel talks about IA’s inherent tension with Agile).

Meanwhile, in 2012 GOV.UK launched. I was alerted by Stephen Gill to an interesting quirk of the alpha of GOV.UK: it had no navigation at all and was search only. While the team quickly learned from feedback that they needed at least some navigation, one commenter noted more generally noted a lack of discussion about structuring information and navigation. Imagine if they had been brought in as Leisa Reichelt was for user research… I mention this as at least one startup founder has noted that the way an organisation has set up sets up foundations that are hard to shake. (EDIT: since writing this I've been told that this was a deliberate counter to Directgov—which I can't speak to as it was before my time—and hopefully the provocation helped challenge existing thinking. That's fair, however this mission probably changed firstly after transactions were made a part of GOV.UK, and then BusinessLink and other more specialist information was also brought onto GOV.UK.)

Extending from the website to the rest of government, in 2013 the Government Digital Service (GDS) rolled out a framework where GDS controlled the GOV.UK website (a kind of hub), and UK departments built the digital forms for getting and giving information (spokes branded as GOV.UK but managed by departments). Much of the push was to move away from all-or-nothing ‘big bang’ launches and instead move towards faster feedback loops based on work tested with real users. To keep quality assurance up, GDS also had spend control for letting services go live or even procure development teams. (In 2021 this moved to the newly created Central Digital and Data Office (CDDO) but that's another story.)

The original explanation of the Service Manual in 2013. Original version on archive.org

This was a good thing! However, this was more directly translatable for simple forms. Even when spend controls brought in colleague facing services in 2015, and service design later took more of a role in GOV.UK services from 2015 on, there was still little available about structuring information for repeat use in public facing or colleague facing services. 

GDS also brought in job descriptions for user-centred design disciplines, which was also good (before that designers apparently had to be told to ignore that their job title was ‘policy officer’ or similar). However, to date it has no clear description of information architecture as a skill let alone who does it (is it the content strategist? If a team didn’t have a content strategist, which role between the service designer, content designer, or interaction designer does it?) which is unfortunate as ever more departments commit to creating single portals (career options) or service-scoped accounts (driving, benefits and so on).

To date, there also isn’t much about information architecture in the GOV.UK Service Manual or even really in blog posts (apart from a couple in 2021).

Back in 2014 (exactly 10 years ago), Christina Wodtke (she of the book that introduced me to the Winchester Mystery House) also wrote about the fall and hopeful rebirth of information architecture….

Mid-2010s onwards: more design hype, fewer professional organisations

As someone in a workplace that does not have security permissions to use Figma, I can tell you that from the outside the fervour around the tool sometimes seems a bit much. (Though this was also what people said about Invision and Sketch in the mid 2010s, so maybe perhaps be careful what tool you hitch your hopes to).

More generally, there is a lot more of a focus on people switching to service design, content design or interaction design either with basic bootcamp training on user-centred design or sometimes even nothing at all (I have seen a lot of developers switch straight to UX with seemingly no taught instruction). I’m no gatekeeper, I understand that people need entries that don’t require several years of study. (It perhaps doesn’t help that some of the early information architects had advanced study in library science to lean upon). However, I have noticed that many people parachuting into these roles don’t even know what they don’t know when it comes to information architecture (starting with the term). 

I also  think that information architecture is doing a terrible job in promoting its value. When I went to look for courses online, I struggled to find good examples from known forums like Udemy or Coursera. There are courses for those that know (for example from Abby Covert and Mags Hanley) but few really visible (apart from new riffs like Object-oriented UX, more on that in a moment). I suspect that some of this may have also come down to some bad timing, such as:

As information architecture is important but decidedly unsexy in comparison to high fidelity prototyping as done in Figma, without visibility through courses and conferences, I think that it’s harder for people to stumble upon it and know its value. 

It does mean that, to echo Andy and Dan’s observations,  that I increasingly see designers start designing complex websites or repeat-use services by going straight to high fidelity screens. To me, this suggests that they haven’t been told about the need to consider structure, and how to make tradeoffs on different tasks serving different user groups (to use an information architecture analogy, like designing a flow for a supermarket or shopping mall). I’ve also come to feel that if I can’t find a sitemap for an existing repeat use service it’s likely no one else thought about structure from a user’s perspective either.

Maybe we’re due for an information architecture renaissance

As many services move from the agile analogy of going from building a skateboard to a bike (or even car), maybe this is the right time to revive awareness of information architecture.

I am particularly excited by the Object-oriented UX work being promoted by Sophia Prater, as it not only builds on existing methodologies but has extra tools (ORCA) to help with screen design. 

There is also another tool that languished for a while before having a second wave of popularity—artificial intelligence (AI). Given that there is a saying going around of “no AI without IA”, and discussions that Responsible AI is an IA Skillset, maybe AI and the need for structured information may help a new interest in information architecture.

If you are designing anything more than a simple flow

In the meanwhile, for anyone designing flows with any amount of complexity (such as things that have repeat use), please have a look at information architecture as a discipline. 

There isn’t much documented about doing agile information architecture, but as the excellent Dean Vipond of the NHS app explains, it is possible to do so with :

  1. Top tasks and user research about things that people can’t find
  2. Initial structure (for example using a common visual vocabulary there are also some other examples in the resources) 
  3. Tree tests to validate - there are good tools for this like Optimal Workshop’s Treejack  

Do also look at the ORCA model (object, relationships, calls-to-action, and attributes) for designing more complex screens - I think that this is also able to be done within agile cadences. 

And if you’re feeling curious, look at the resources I mention. 

More on information architecture

Here are a few resources that I’d recommend.

Getting started

More detailed techniques

Aside from some of the courses I have mentioned, here are a few more: 

  • Information Architecture, (4th Edition), Louis Rosenfeld, Peter Morville, Jorge Arango (2015) - affectionately known as ‘the polar bear book’, this is basically the bible for IA, clocking in at 400 pages. It has a lot on search and facets. 
  • Design by Definition, Elizabeth McGuane (2023). This has a lot of standard information architecture stuff but also does more of a deep dive into terms and models.
  • Card Sorting: Designing Usable Categories, Donna Spencer (2009). One of the original books on the topic, it is really the primer for doing the technique effectively, with tips such as doing only 30-100 cards (sweet spot for grouping) and being intentional about not mixing level of detail or feature vs content. It’s also a bit of a time capsule of card sorting beginning to be possible with online tools, and also has a lot of references to the IA Institute (RIP).
  • Information Architecture: Blueprints for the Web, Christina Wodtke (2002) - this is long out of print, and so there are some entertaining references to CDs and the like, however this is still solid, also with examples of sitemaps and spreadsheets. For me it was worth it if only for the house story. (Update: there is a second edition from 2009 co-written with Austin Orella which I can't talk to but is available on O'Reilly)