Designing content that caters to diverse levels of expertise
I recently worked on a local government website with a common focus: how do we ensure our audience can find and identify what they’re looking for?
This article covers some of the challenges and approaches surrounding this focus, with particular attention to designing for the audience’s diverse levels of familiarity and expertise with the website’s underlying subject.
Local governments do many different things: from parking permits, development applications and waste management to parks, libraries and pet registration.
Local government also serves a broad audience — not just people who live in the local area, but also businesses and professionals who operate within it and visitors and tourists passing through. We don’t design for individuals; we design for whole user populations, with different levels of familiarity and expertise, in endless ever-shifting contexts. This reminder is especially crucial for government products, given the audience has no choice but to use them.
It’s this aspect of ‘different levels of familiarity and expertise’ that this article focuses on. How do we label and describe processes surrounding things like public domain works and development applications when the audience includes someone who’s done it many times before or does it as part of their job, alongside someone who’s renovating their home for the first time?
Novices and experts think differently
The concept of basic-level categorisation helps articulate how a novice may think and talk about something in a different way to someone who has studied and built familiarity with it (an ‘expert’). The concept can be traced back to Roger Brown’s paper How Shall a Thing Be Called? (1958) and has been explored further by Brown and many others since.
The basic level refers to a particular level in a category hierarchy. This level is where natural thought commonly takes place and where things are most readily described. For example, when I see a furry thing with four legs and a tail being walked down the street, I’m likely to refer to it as a “dog”. Generalising from here, I may refer to it as a “pet”. I can also get more specific by referring to it as a “golden retriever”. In this context, the term “dog” sits at the basic level, and for many, it would be the most readily available descriptor for this thing. We can describe the same phenomenon for many other subjects. I call that thing sticking out the ground a “tree” rather than a “plant” or a “Moreton Bay Fig”, and that thing travelling on the highway is a “car” rather than a “vehicle” or a “Ford Falcon”.
Basic-level categorisation asserts that we don’t start at the top of a conceptual hierarchy and work our way down — we ‘think in the middle’. This is because the basic level is where the majority of salient and useful information exists. When we generalise upwards to the superordinate level, we lose a lot of that information. When we specify downwards to the subordinate level, we gain only a little. At these other levels, there seem to be “no characterising actions”. They are “achievements of the imagination” (Brown, 1965).
Expertise interferes with the basic level
As you may already suspect, ‘thinking in the middle’ is not as universal or stable as I’ve suggested.
In their paper Familiarity effects on categorisation levels of faces and objects (2009), David Anaki and Shlomo Bentin describe several scenarios that “induce a shift in the default level of object categorisation from basic to individual level”. That is, being more inclined to refer to that animal with four legs as a “golden retriever” rather than a “dog”. Anaki and Bentin call this subordinate-shift. Their list of scenarios include:
- Being an expert on a subject. Experts are likely to use more specific (subordinate) terms when describing something.
- Having familiarity with something, and this familiarity not necessarily stemming from an expert understanding of its broader domain.
- Referring to atypical members of a category. For example, calling those waddling things in Antarctica “penguins” rather than “birds”.
What does this all mean for labelling local government content?
Let’s assume that our inclination to refer to and naturally think of things using basic-level categories translates to how we identify and search for content online. In other words, if we define content instances at a level of specificity that matches how people typically think about them, then that content will have a better chance of being found and accurately assessed by a broad audience.
To do this, however, we’ll need to overcome two challenges:
- Identifying the basic level: Determining where the basic level sits isn’t so straightforward for non-object domains. For example, how do we identify the right level of specificity and terms for content about development applications?
- Writing labels that are suitable for everyone: Without duplicating content, implementing excessively complex personalisation, or writing long and hard-to-scan titles, how do we label content so that both experts and novices can readily find and identify it?
1. Identifying basic-level categories within the government domain
Dogs, trees and cars all have relatable physical forms. This tangibility is a big part of why they are ubiquitous basic-level exemplars of their respective domains. Brown and others have identified this ‘relatable physicality’ and many other properties that characterise categories that share this success:
- Languages often have simpler names for things at the basic level. “Dog”, “cat”, “tree”, and “car” are a few English examples.
- Concepts at this level have easily distinguishable morphological characteristics. For example, a dog is typically furry, has four legs and has a tail. Most dogs don’t diverge from this when we go down to subordinate breeds.
- It is the level of distinctive actions. For example, I might make a wafting, smelling motion when describing a flower. Going down to a specific species of flower won’t typically add meaningful distinctions to this action.
The problem is, many of these properties are not available to government concepts. What is a distinguishable morphological characteristic of supplying an annual fire safety statement? Or a distinctive action of a late-night trading strategy?
Tactic 1: Explicitly design the structure of intangible concepts
When developing the content for a council’s website, the project team and I modelled the local government domain. During this exercise, we encountered several intangible, enigmatic concepts.
Rather than translate these concepts into a ‘general’ template, we opted to create a new content type with its own set of attributes. We wanted to do this because we frequently observed users interpreting this content as lofty and approaching it with cynicism. Perhaps by building a robust structure around this content that acknowledged this attitude, we could bolster its salience. If successful, we’d bring the content closer to the tangible qualities of basic-level categories found in surrounding domains. In turn, this would improve its comprehension, highlight its relevance, and tackle the audience’s cynicism head-on.
The result was the Focus content type, defined as a bridge between a strategy and the projects and programs that make up its delivery. Its usage guidelines encouraged authors to write about an instance’s objectives and be explicit about why it’s worth attention and ratepayers’ investment. Backing this up were relationships to projects and initiatives undertaken by the council, building tangibility around organisational focuses like social sustainability, the night-time economy and community engagement. This moved them away from their perennial inclination to be perceived as lofty or panderous.
With all this said, the Focus content type didn’t identify the content’s basic-level category; however, we can still link the tactic back to the framework because it accessed the qualities of basic-level categories by imitating their characteristics. We achieved this through the explicit design of the content type’s structure, which emphasised relationships to tangible projects and programs.
Tactic 2: Match the specificity of content to how users talk about it
Talking to users about their experiences related to your content’s domain helps you understand how they articulate their information needs. This process will reveal their preferred terms and the level of specificity in which they naturally think of your content, i.e. their basic-level categories for that particular domain.
We constantly interact with local government services, programs, and spaces, so in my case, finding a relevant story to discuss with users was easy.
These were not ‘evaluative’ conversations. The conversation needed to feel natural to allow tacit understanding to come to the surface. If the evaluation of a prototype were the mainline conversation, we would distract the user from articulating themselves naturally.
The key was identifying an appropriate underlying activity. For example, during a sprint focused on content like sports competitions, youth programs and grants, we recruited users based on recent experiences participating in these sports competitions, registering their kids for a youth program, or applying for a grant. During the interviews, it was then simply a case of watching them do these activities and talking to them about it (i.e. contextual enquiry), or if that wasn’t possible, having them recall it in detail: what happened in the order that it happened, while leveraging cues available in the environment: be that physical space, or calendars, texts, and other trails to ground the discussion in the real event.
This approach may feel distant from the thing you’re designing, but it far better reflects the reality in which users discover your content. After all, people don’t arrive on a website and start navigating – they’re already navigating, every moment of every day, using the resources they can find and understand to satisfy a broader intent.
Specificity is identified through analysis of the user’s vocabulary and how they accomplish their underlying intent
Looking to the same example of content about things like sports competitions, youth programs and grants, one person we spoke to talked about their role as their office’s social events coordinator. This person would need to determine suitable activities that people would want to participate in. They would do this by selecting group activities within walking distance in their assigned budget and then sending a survey to staff to determine which ones resonated.
We extracted a few insights from their storytelling:
- Location and cost were both non-compensatory: sports competitions had to be close enough to be a feasible lunchtime or before/after-work activity. They also had to be within a specific budget.
- Sports competitions had to be easily assessable: in our participant’s work practice, activities had to compete for attention in a long list of options. If they demanded further investigation because of ambiguous details, they’d likely get dismissed.
- They had to appeal to a hesitant audience: the success of selected programs was measured by engagement. Not everyone surveyed was a seasoned volleyball or indoor soccer player — reassuring details like what equipment is provided and if any prior experience is needed would help abate uncertainty.
Considering these observations, we landed on defining an instance of a sports competition by its location and sport facets; for example, “Volleyball competitions at XYZ Centre”. This was also supported by our analysis of the specificity of how interviewees typically articulated concepts in this domain: “we played volleyball at XYZ Centre”; not more specifically “we played beginners volleyball at XYZ Centre”, nor more generally “we played sport at XYZ Centre”.
If we generalised and only included a sports competition's location, it would be too ambiguous to be easily assessed. Although the facet of ‘time’ is an important aspect for many, most sports competitions had sessions in the morning, at lunch and in the evening, so we left it as prominent content on each sports competition’s page instead. The same can be said for ‘grade’: in the context of a social event coordinator’s work practice, this would only risk dissuading a hesitant audience from considering it unless there was more information; so, as we did with ‘time’, we left it out of the title.
Naturally, it would be best if you made these decisions by consolidating observations from multiple sources, but this example still communicates the premise. Listen to how users articulate their information needs in the target domain and complement this with extracted requirements uncovered by analysing your users’ work practices: how they accomplish their intent, who is involved and what distinguishing characteristics are important.
2. Writing labels that are suitable for everyone
As Anaki and Bentin’s concept of subordinate-shift describes, basic-level categories are not where natural thought occurs for everyone, nor is it stable across all domains and instances. Experts express themselves in more specific terms than the typical user; furthermore, expertise can emerge quickly – a sense of familiarity is enough to induce a shift to more specialised language.
This poses a problem for describing content on the web. The digital world’s “wealth of information creates a poverty of attention” (Simon, 1971), which brings with it a pervading tendency to scan and bounce around for good enough. The long titles and descriptions necessary to capture both novices and experts' attention won’t play nice with this behaviour.
The following tactics describe methods I’ve used to capture novice and expert ‘trigger words’, plus guidelines produced for local government to label and describe the content in a way that acknowledges subordinate-shift.
Free-listing and performative closed card sorting
I frequently use a card sorting format that attempts to simulate subordinate-shift. In this format, a closed card sort is bookended by asking the participants to free-list (I say “term”, you think what?) against the same categories presented in the card sort.
Using this structure, the card sort is ‘performative’ in the sense that its primary purpose is not to see where content is sorted (although that is still of interest), but rather to cultivate familiarity with the facilitator’s proposed categories and then observe differences in the participant’s free-listed exemplars before and after sorting.
Using this format, I’ve found the pre-sort free-listing exercise uncovers numerous potential trigger words for information scent attractive to novices. Post-sort, the same free-listing exercise reveals exemplars that are often more specific, have fewer family resemblances, but still great potential salience.
In the government project’s case, the harvested exemplars informed category labels and descriptions, which was especially important given the superordinate categories typically demanded by the domain’s broad and disparate content.
Titles are for novices; descriptions are for experts
In their paper Patterns of information seeking on the Web: a qualitative study of domain expertise and Web expertise (2003), Jenkins, Corritore and Wiedenbeck describe traits of a domain expert’s online search behaviour. Here are two of them:
- The use of advanced terminology: domain experts rely on their extensive vocabulary to construct more topical queries than domain novices.
- Effective evaluation: domain experts are faster than novices at determining a page or search result’s relevance.
There are clear similarities between the use of advanced terminology and Anaki and Bentin’s concept of subordinate-shift. Additionally, Jenkins et al.’s claim that domain experts can evaluate items' relevance quickly is another important consideration to help us determine where we should place expert and novice vocabulary within a search result or page.
On the local government website I worked on, the Service content type best demonstrates the application of Jenkins et al.’s observed traits. The Service content type represents things like parking permits, waste collection, development applications and property certificates. Instances are collectively defined as “something that someone does to fulfil an intent”. Not all that specific, but a helpful reminder to content authors that users just want to get their parking permit or find their bin day and get out – extraneous details like parking schemes and environmental policies are better off being housed on a separate page, linked to, and kept out of the way.
Local governments serve both novices and experts in comparable volumes. As such, these services need to be identifiable and match the vocabulary of a user’s search for a broad spectrum of familiarity and expertise.
To address this, we decided to reserve titles for novices and descriptions for experts. For example, titling a service “Get permission to build or renovate”, as opposed to “Lodge a Development Application (DA)” or “Apply for a Complying Development Certificate (CDC)”, which include specialised terms that may be unfamiliar to the user and their intent.
Descriptions are then used to surface an expert's vocabulary to build information scent for audiences with greater familiarity with the service. For example, the service “Get permission to build or renovate” may use its description field to surface terms that accredited certifiers and construction workers typically use, such as “CDC”, “DA”, and so on.
We can enhance the effectiveness of this approach further with hit highlighting. Although the description is less prominent than the title, highlighting searched terms will speed up a user’s assessment of a result’s relevance without detracting from the novice vocabulary found in the title.
GOV.UK utilises a similar approach, as expressed in their internal axiom “Good services are verbs, bad services are nouns”. In the world of GOV.UK, “DA” is a ‘government noun’, whereas the underlying intent (getting permission to build, renovating your kitchen) is the ‘verb’:
The trouble with names like these [‘government nouns’] is that you need to be introduced to them before you can use them, meaning that part of ‘doing a thing’ means learning what government calls the thing you’re trying to do.
The strategy is robust, but I wouldn’t abandon the ‘government noun’ altogether. For the website I worked on, an analysis of search logs clearly showed terms like “DA”, “CDC”, and other expert terminology included among common queries. It was this observation that brought the project team to declaring descriptions as expert territory.
Here’s a summary of what was covered:
- Novices and experts think differently. This difference presents itself in the specificity of someone’s vocabulary, extending to how they search for information.
- The way we think is unstable. People can become familiar with a subject quickly, and this familiarity is often enough to induce a shift in how they think and search.
- You can inform the specificity of your content by listening to users. The way people naturally talk about their experiences with the underlying activities your content represents is rich with this information.
- If you‘re dealing with less tangible concepts, try explicitly designing its structure. By doing this, you might discover how to express the concept in a way that’s more useful to its audience.
- Allow content authors to write for both novices and experts. This could be by designing an interface that accommodates longer titles and descriptions. You could also provide clear guidance on how titles, descriptions, and other fields should be populated.
These points chip away at how we search for and make sense of information, but it’s only a small part of an everlasting challenge. Remember that, without exception:
- stability in any domain is an illusion
- all categories reflect a subjective reality
- every task is always situated in a broader context.
These truths ensure the continued proliferation of including “make the content easy to find and understand” in every brief, so please share your strategies in tackling this complex design challenge, and let me know what you thought of the ones presented here.