Editorial
Raise your hand if this sounds familiar: your intranet is showing its age, but you don’t have the budget to deliver the upgrades you’d like.
Instead of viewing budgetary restrictions as a blocker, you should view it as just another guideline for what you can and can’t do right now. And there are a number of things you can do without incurring too many costs.
At the heart of any improvements should be a focus on better aligning the range of internal digital communication, services, and collaboration with the needs and expectations of employees.
While there are many options during this time, such as stakeholder interviews to learn about challenges around digital working, revisiting your vision and mission for your digital work environment given all the changes in the digital landscape, and developing team agreements about which tasks to accomplish where, for the purposes of this article I’d like to dig into improving the navigation structure of your intranet.
Improving Your Intranet Navigation Structure
It’s a good practice to revisit your intranet’s navigation structure from time to time as it expands, items are added, and content shuffles around.
You can do so in five steps:
- Gather input
- Card sorting
- Create draft navigation design
- Tree test
- Finalize navigation design
Of course, you then have to process the result in your intranet, but that is beyond the scope of this piece.
Gather Input
To improve the navigation of your intranet, you should start with the structure of your current solution. Collect all the items in the menu, both at the highest and deeper levels, and list them in alphabetical order in a spreadsheet.
Compare this list with any unrealized plans that you have made in recent years for innovations to your intranet. Identify the topics that may need a place on the intranet in the future.
Also, check your notes from any intranet-related conversations you’ve recently had with stakeholders. Do any items stand out as future additions? If you've done research on employee needs more or less recently, use that information as well.
Your last place to look is at the structure of other digital systems for HR or Facility information to find any topics they cover that may be relevant to the intranet. Copy these topics into the list to better inform and refer employees to these services.
Your current intranet navigation likely has a list of dozens, if not hundreds, of topics already in place.
That is far too many for the next step, card sorting, so you'll have to prune here as well. You can, of course, delete duplicate topics, merge similar or close themes, and delete the items that go into too much detail. Ideally, that will leave you with between 60 and 100 items. Each of these items should:
- Be clear and unambiguous
- Be unique and not overlap with other items
- Be short and to the point, but not too short that they become unclear
This can take quite some time, but you haven't spent a dollar yet. You may need several iterations with a group of colleagues, but that is also very normal. A practical tip: save any major changes you make as a new tab in your spreadsheet so you can refer back to previous versions. Microsoft Excel and Google Spreadsheets also save versions, but finding something in them is always quite a puzzle for me.
Once you have a version that you are happy with, you are ready for the next step: card sorting.
Related Article: Does Your Digital Workplace Design Help Employees Get Work Done?
Card Sorting
Card sorting is a golden oldie. In the past, it was literally what the name suggests: you arrange physical cards in a way that makes sense. After employees finish organizing the cards in a logical way, you’d give each category they created an appropriate title.
The advantage of physical card sorting is you can do it with a small group, where you invite them to discuss and coordinate in order to arrive at a shared structure.
Individual card sorting is just as workable: each participant groups the cards in their own way, and afterwards you process the results (more on that later). Digital card sorting comes in here. I've been a satisfied customer of Optimal Workshop for years, but there are several alternatives, including Maze, UserZoom, and UXMetrics. You can also do card sorting with a digital whiteboard such as Mural or Miro, but that requires manual processing.
Card sorting can be done in three ways:
- Open: Participants group the cards and decide for themselves which categories they create and which titles they give them.
- Closed: Participants group the cards into a preconceived set of categories. So, people can't add new categories.
- Hybrid: you might have guessed it: Participants group the cards into predetermined categories, but can also add new categories.
For internal digital platforms, I always use the hybrid variant and always use the same predefined categories:
- About me — information about my employment
- Find a colleague and collaborate — for example, profile pages and social interaction
- About [organization] — background and policy information
- News — current events and events
These categories come from a study Gerry McGovern conducted of 55 intranets around the world. The exact wording may vary and you may want to add one or two from McGovern's research, but the thrust is often the same.
Conference
Sep
16
HR Tech Conference Las Vegas 2025
Register
Conference
Oct
27
Gartner HR Symposium/Xpo Orlando 2025
Register
Depending on the digital tool you choose, you can then set up your test. Enter the tickets/items, create the categories, and add any segment questions (role/function, location, etc.). The latter can help uncover similarities and differences in the responses in order to potentially offer segmented navigation, but more often I use the responses to demonstrate a broad group of colleagues were involved in the exercise.
Aim to have at minimum 40 participants in this exercise as that’s the point when some patterns emerge. An organization of 1000 employees should involve around 100-150 participants (i.e. 10% - 15%).
The tools often can show you the results in different ways. Below are two examples: on the left is a standardization grid and on the right is a similarity matrix.
Create Draft Navigation Design
Card sorting has one major drawback, if you can call it that: after the test round, you don't have a ready-made navigation structure. Depending on your tool, you have all kinds of insights into the patterns, which cards are put together the most, in which category, and what title that category gets. So you have to process those insights.
The advantage of digital card sorting is it reveals patterns in employee responses. But you have to discover those patterns, analyze them, and incorporate them into a draft navigation structure or menu design.
In my experience, this remains manual work. Sometimes you have to make choices if the results of the card sorting are unclear. But that's OK, because in the next step you test your choices. It is also wise to discuss a draft structure with various stakeholders to gain support for the results of the card sorting and therefore for the input of the next step, the tree test.
When the draft navigation is ready for the test, I always add 'Search' to the tree structure. That way you get better insight into how people perform the assignments you give in the tree test. If you don't add the Search option to the structure, a) you'll miss out on that insight into the test results, b) you run the risk that people will select other items that they don't really consider seriously, or c) they'll skip the assignment and you'll miss out on data.
Related Article: Why Your Internal Content Needs a Taxonomy
Tree Test
The second round of testing is the tree test, or another digital survey where you test the tree structure of your navigation with employees. I use Optimal Workshop for this also, but alternatives are UXtweak, UserZoom, and Maze.
Three things are important in a tree test:
- The tree structure, or your concept navigation design, is the result of the previous steps.
- Possible segmentation questions: I usually use the same questions as for card sorting.
- The assignments on the basis of which you test the structure: I will discuss them in more detail.
Earlier, I wrote about the importance of keeping a clear eye on the needs and expectations of employees. I've been using task identification for that for years. Task identification results in a list of four to six 'tasks' — which can be activities or actions, but also functionalities or types of information — that receive about 25% of the total number of points from study participants. Explaining it would take a bit too long to go into detail here, but a practical example is from an intranet at a hospital, which might show the following priorities:
- Patient record: 9.3%
- Find a colleague: 9.0%
- Files: 5.7%
- Work schedule: 5.0%
- (Medical) procedures: 4.9%
- News: 3.7%
- Education: 3.4%
- Salary: 3.0%
- Leave off: 2.6%
- Work-life balance: 2.4%
You can see that the three highest-scoring items together received 24% of all points, and the top 10 received just under half (48.9%). With these 'tasks' you can formulate assignments for the tree test that are most important from the point of view of employees to be able to carry out on the intranet or within the digital work environment. Now, the phrases above are too concise as an assignment — they are also a bit more extensive in the task identification itself — but they offer guidance.
- You want to make a note of your patient's treatment plan. This is a concrete assignment based on the first task, with basically only one correct result in the navigation structure.
- You are looking for a colleague who can help you with a question or project. This is a specification of the second task with one correct result.
- You want to consult the antibiotic guidelines. This assignment is based on the fifth task from the list above, and again there is one correct result.
The list of tasks from the survey is a very good guide, but sometimes important assignments do not score high, but still serve an organizational interest.
I usually give between eight and 12 assignments for a tree test. These are typically the highest-scoring tasks from the task identification, but sometimes include other assignments.
As with card sorting, you don't need hundreds of respondents for the tree test, but patterns will emerge with about 40 people. Again, you should aim for 10%-15% of the employee population so you can defend the reliability of the results. I usually give a week for the responses.
The result of the tree test shows roughly two things:
- The results of all assignments overall, e.g. 76% success. This means that more than three-quarters of the respondents were able to find the right answer.
- The results of the individual assignments. For each assignment, you can see how many people found the right answer, and how many people choose alternative places in the menu.
The use of alternate paths doesn’t necessarily mean it’s a bad result, at times it just suggests an unanticipated nuance which will require an adjustment in your intranet implementation.
An important caveat to note: the tree test does not provide feedback to the participants on whether or not they have found the right location. For most respondents, a tree test is a one-time test, so there is no learning effect. Respondents indicate what they think is the right place, receive no feedback there, and therefore cannot, either immediately, or next time, choose an alternative.
Of course, the latter is the case on a new intranet. This means that we should not only look at the 'flat' response, but also at the (alternative) click paths. And you can take into account the learning capacity of users of the new or renewed intranet.
Related Article: 5 Intranet Trends Redefining Employee Experience in 2024
Finalize Navigation Design
This brings us to the final step of optimizing your navigation design, namely incorporating the results of the tree test into the concept structure. For this stage, you comb through the results of the various assignments and adjust the structure where necessary. You can do two things:
- You have to adjust the location of your content or function if a majority of respondents expect to find the answer to the assignment in a different place than in your draft.
- You can link from that other place in your structure to what the majority thinks is the right place (and what has also emerged from the card sorting) if a large minority of respondents expect the answer elsewhere.
With these adjustments, you can then go live. And then the litmus test begins, because it’s only in practice when you’ll see if the changes have the desired result, if people’s mastery of the navigation gets better and if the adjustment is really an improvement.
Ultimately, while this kind of work doesn’t require great financial costs, it does take a great deal of effort and time, and organizations will need to decide whether to make this investment.
Learn how you can join our contributor community.
