Dualo chevron
Back to blog
Taxonomies: the make-or-break of an insights repository
Research excellence

Taxonomies: the make-or-break of an insights repository

Knowledge is only as powerful as our ability to make use of it. Taxonomies make information accessible and can inspire the exchange of insights across teams and organisations - but what does it take to actually build one?

Dan Robins
November 11, 2021

By definition, taxonomies are systems used to classify and organise data. The earliest taxonomies date back to the origins of human language. From the animal kingdom to a digital sitemap, whenever you see the categorisation of things, this is essentially a form of taxonomy.

Taxonomies are certainly nothing new, but we continue to find new ways use them for one simple reason:

Taxonomies make information accessible. 

In the context of user research, building and maintaining a collection of tags (or in other words, a taxonomy), is essential in order to achieve user research excellence. At Dualo, we firmly believe that a strong taxonomy is the backbone of every great knowledge repository. 


In this article, we’ll look at:

- Why building a taxonomy for the outputs of your user research is so important

- The cost of not having a strong taxonomy

- Who should build and manage taxonomies

- How taxonomies are structured, and 

- A practical guide to building your own taxonomy

Why are taxonomies so important when building insight repositories?


Your organisation likely has many touchpoints and a wide variety of channels for obtaining knowledge. This might be through direct customer feedback, quantitative and qualitative analysis, or capturing technology and industry insights.


This knowledge is often valuable beyond the scope of a single study or context. In fact, product insights that relate to human nature and behaviour are likely relevant for entire organisations, and can remain true for years. 


But all of that knowledge is meaningless unless you can make sense of it. And the volume of available data that organisations can accumulate can often be overwhelming.


It’s therefore important that the outputs of user research are consolidated and organised in a way that’s accessible to wider teams and stakeholders. After all, insights are only as powerful as our ability to make use of them.


A taxonomy allows you to divide and group large collections of knowledge into more manageable themes. They help to visualise and organise your data.


Depending on the goals of your insights repository, a well structured taxonomy can:

- Efficiently and effectively filter existing knowledge 
- Help identify patterns in your data and surface new insights for further consideration
- Provide guidance to consistently and holistically categorise new data
- Provide strategic direction to stakeholders and wider teams

The cost of not having a strong taxonomy 


“Those who cannot remember the past are condemned to repeat it”

George Santayana

If the benefits of having a taxonomy aren’t compelling enough, the cost of not having one is enough to make leaders stand up and take action in organisations large and small.

Research by nature is exploratory. Taxonomies structure and categorise information in a way that allows us to explore a particular subject or theme - bypassing the need for a specific search term or a preconceived idea of what exactly you're looking for.

Consider the scenario of onboarding onto a new team or project. The insights your organisation holds are likely spread across multiple previous studies. Insights specific to your needs may have been discovered in much broader contexts and remain buried deep within larger reports.There’s no doubt it would be helpful to have access to previous explorations and existing knowledge relevant to the area you’ll be working - but without the ability to filter knowledge, it’s difficult to know where to look. 


Many teams and individual researchers we’ve spoken with find themselves in this position more regularly than leadership often realises. In fact, according to a study conducted by Qatalog and Cornell university, employees waste around 59 minutes a day searching for knowledge.


It’s very common for teams to therefore start from scratch and create new silos. A recent IDC study found that data and research professionals lose a further 20% of their time duplicating research.


So aside from the significant opportunity cost caused by a lack of structured, inaccessible data, there are significant hard costs too. Considering the average researcher salary in the UK, searching for knowledge and repeating work costs organisations £16,250 per researcher, every year.


Searching for knowledge and repeating work costs organisations £16,250 per researcher, every year.

Who should build your organisation’s taxonomy?


Research professionals within customer-centric organisations have increasingly become the orchestrators of knowledge, responsible for storing data and insights in a retrievable way. They will likely lead the charge as they already have the clearest understanding of how knowledge should be structured. 


That said, building a taxonomy for a knowledge repository should be a collaborative process. When initiating the process, it’s important to consider not just teams who will be contributing to the repository, but also the teams and individuals that will be consuming data from it. 


Taxonomies should be built with multiple teams and disciplines in mind. Using universally understandable language for tags and themes encourages wider teams and stakeholders to self-serve knowledge and build an understanding of how data is structured across your organisation. It’s often helpful to therefore involve representatives from these teams, at the very least when reviewing or refining your taxonomy. 


Local or global? 

A question we often get asked is “where do our local taxonomies fit into all of this?”. By ‘local’ what people are often referring to is taxonomies specifically built for teams, departments or projects (rather than representing a more global, multi-team view). 


Whilst we’ve seen localised taxonomies work, our view is that (for the most part), organisations should work towards one minimum global taxonomy. Why silo your data and allow inconsistencies as well as duplication to occur when you can build one powerful and all encompassing system?


How is a taxonomy structured? 


Since no organisation is the same, there’s no ‘one size fits all’ solution to building a taxonomy. And because data can tell many different stories, it’s important to establish what you want to learn from the data before you decide how to structure it. 


Once you’ve established what you’re hoping to achieve by building your taxonomy, you’ll need to define its scope. Are you building a multi-disciplinary taxonomy to classify your data at an organisational level? Or is the scope narrowed to only ‘qualitative research’ or ‘industry insights’?


Scope plays a major role in the granularity of categories; in qualitative research tools, tags might be used to categorise raw data, observations and singular sentences from a user interview. In a multi-disciplinary repository, tags are more likely to span across multiple themes, and cover broader products and teams. 


One of the key jobs when building a taxonomy is to establish a set of overarching themes. Themes are essentially patterns in the data. 


Generally speaking, insight repositories are most effective when built using a hierarchical taxonomy, which enables us to create these sub-categories or ‘themes’. 

A ‘hierarchical taxonomy’

For many reading this post, a hierarchical taxonomy will likely have a strong resemblance to a UX affinity map/diagram. 


And for good reason, the process of building taxonomies is very similar to a card sorting or affinity mapping exercise. Both use similar methods to organise research findings into overarching themes.


Below are some common themes we’ve seen teams include in their taxonomy:


- Product / Feature / Touch point

- Sentiment / Experience / Pain point / NPS

- Segment / Geography / Audience type / Persona

- Need / Goal / Motivation / JTBD

- Device / Browser / Technology 

- Research Type / Method  / Researcher / Source / Date

- Team / Tribe / Department / Discipline 

- Scale / Frequency / Severity / Business impact / goals

- Industry / sector / Competitor / Proposition


Remember that when it comes to tagging, not all themes will be relevant to every piece of research.

Is there a maximum recommended number of tags or themes?

There is no set amount of tags or themes, nor a maximum number to stick to when building a taxonomy. We find it helpful to follow the ‘just enough’ principle. Building a taxonomy is as much an art as it is a science, so you’ll need to experiment with what’s right for your organisation and use case.


Too many tags can be overwhelming and slow down researchers uploading data. Too few tags diminish your ability to usefully filter knowledge in specific areas. Generally speaking, it’s advisable to build broad and shallow rather than narrow and deep. If your taxonomy has too many levels, potentially useful themes and patterns could end up getting buried. 


A practical guide to building your own taxonomy 


For experienced research teams, experimenting with methods for building and maintaining a taxonomy is common practice. We’ve worked with many teams and seen several techniques work particularly well. We've developed a workshop inspired by some of these practices and structured around the affinity diagramming method. The workshop is flexible and intended only as a guide. The workshop takes roughly 1.5-2 hours, we hope you find it useful.


Taxonomy workshop:

1. Establish the desired outcomes

What is the scope of the taxonomy?

When do we want to begin using the new taxonomy?

2. Analyse sample data

Depending on the scope of your taxonomy, we’ve found that a mixture of data types works well. Try to bring between 5-10 sources to deconstruct. Sample data could range from existing studies and reports, to strategic insights, analytics reports, or UX artefacts - whatever is relevant to your scope, goals, and organisation. 

3. Establish an initial set of tags from the sample data 

This doesn’t have to be perfect or exhaustive. What categories jump out when looking at each piece of sample data? 

4. Create sample themes

Consider the themes and how you want to categorise your data by looking at the initial set of tags you've established. These can shape, grow and merge as time goes on. Consider creating a few category names to help get the sorting started. Some common themes for insight repositories are listed above.

5. Add a “?” theme

Adding a miscellaneous category is helpful so as not to get stuck on tags that are less easily categorised. Tags in this category can be revisited later with a more holistic view of the data. 

6. Begin sorting & identifying themes in the data

1. Choose a tag at random

2. Look for a category that makes sense 

3. Move the tag to sit within that category, or create a new one. If no obvious category name arises, place it in the “?” category for discussion later.

7. Assess your categories 

Do your themes and categorisations make sense? Are there any stronger themes that have emerged? Is there any crossover? Can any categories be merged? 

8. Assign colours to each theme

Assigning colours to different themes in the data is helpful for a number of reasons;

1. Contributors can use colour to guide them and ensure multiple themes have been considered when tagging data.

2. Consumers can quickly build visual associations to the themes in your repository, allowing them to visually skim the data and quickly determine if it’s relevant to what they’re looking for.

3. Colours can also be used to signal sentiment. For example, some teams use red for anger, and green or blue for calm.

Assigning colours to different 'themes' or 'categories' in your taxonomy is helpful for a number of reasons.

Following the workshop:

Build your taxonomy and begin indexing your data

If you choose to conduct your taxonomy workshop in-person or outside of the insights repository tool, the next step is to ensure that the tags and themes are built using a tag manager that enables quick and easy setup and maintenance.


Maintain your taxonomy 

Taxonomies are not set in stone, they are living artefacts that are constantly adapting and evolving. It’s important to plan for a learning curve and set reviews with your team on a regular basis. Returning to your taxonomy and spotting new themes in the data is an opportunity to discover new patterns in your user’s behaviours. Consider some common questions when reviewing your taxonomy:


Have any new themes emerged?

Would it be helpful to merge some tags or themes? 

Can we remove any tags or themes? 

Having a taxonomy manager that indicates how many instances a tag is in use is particularly helpful in the review process. Consider merging similar tags that have been used in a small number of instances.

It’s important to be flexible when reviewing and maintaining a taxonomy. The strengths and weaknesses of your taxonomy will become clearer as you tag more data. As time goes on, you’ll most likely need to adjust the structure according to changing processes, scope and requirements.

Tag hover states in Dualo provide a breakdown of the theme that individual tags belong to, as well as how and where it's in use.

Consider who has editing rights to your taxonomy 

Taxonomies take time to build. They allow us to create intricate webs of data, delicately threaded together. Complex data can quickly get tangled up in the wrong hands. To avoid laborious unpicking, ensure that you only give editing rights to people who understand the implications of changing things.


Conclusion

Knowledge is only as powerful as your ability to make use of it. Taxonomies are the backbone of every great insights repository, and one of the most critical jobs to get right when building one. 


Building taxonomies is not just about organising data, it’s about building a holistic view of the knowledge your organisation holds. 


If built and managed correctly, they can make knowledge accessible and inspire the exchange of insights across teams and organisations - enabling teams to elevate to ‘research excellence’ status, and ultimately build better products. Done wrong, they can seriously hamper the ability for researchers to make an impact, and reduce the amount of repeatable value an organisation can realise from research.


We continue to shape our understanding on how teams can ensure they build the best possible taxonomies to maximise the value and impact of their knowledge. We hope that this post proves valuable for anyone looking to build or maintain a research taxonomy.

Happy tagging!

If you found this article useful be sure to grab a free copy of our User research is broken: A guide on how to level up your research operations playbook, available for download here.

References & further reading

https://dovetailapp.com/blog/what-we-learned-creating-tagging-taxonomy/

https://blog.getenjoyhq.com/user-research-taxonomy/ 

https://supermassive.ca/ux-design/building-a-user-centered-taxonomy/

https://www.nngroup.com/articles/affinity-diagram/


About Dualo

Dualo is an insights hub used by digital product teams to get more repeatable value from their user research and insights, so that stakeholders can make informed and timely decisions across the organisation. If you're interested in learning more, please request a demo and a member of our team will be in touch.


ABOUT THE AUTHOR
Dan Robins

I’m a design, UX & strategy lead with a passion for storytelling. Proud member of Dualo’s founding product trio. Always seeking new inspiration.

Insights to your inbox

Join our growing community and be the first to see fresh content.

You're subscribed! Stay tuned for insights to your inbox.
Oops! Something went wrong while submitting the form.
Dualo checkbox

Repo Ops ideas worth stealing

Dualo checkbox

Interviews with leaders

Dualo newsletter signup