As UX Designers & Researchers we collect all this knowledge along the way, most of which is tied to specific projects. The crux though is that the insights can be highly relevant to later projects or even kick off new projects.

My aim is to build a type of “knowledge archive” about our users, encompassing what we know across any type of user (e.g. from potential customers to most engaged users), across any categories of insight (e.g. spanning from qualitative anecdotes to quantitative reports) in any context (e.g. general best practices, exploratory research or project-specific usability insights).

The main question is - the Information Architecture could be quite tricky and seem to be best served by tagging each piece of “insight” in a certain way; what would the best way to do this?

I assume that the following criteria are important towards the success would be -

  • Quick filing of results and tagging
  • Easy filtering and a plethora of options to navigate the insights
  • Extensive collaboration options to be shared across the product team
  • I've removed the request for software here. The question should focus on the actual UX issue, not requests for different pieces of software that will get out of date and leave answers less useful over time. – JonW Oct 15 '14 at 10:50
  • Good shout Jon - sorry about that! – Andreas Wolters Oct 15 '14 at 10:52
  • 1
    This is a really good initiative on your part, and I think you can draw a lot of parallel from the user profiling techniques used in segmentation analysis of consumers in marketing research as well. If I can find more details and reference I'll probably throw in an answer for you. – Michael Lai Oct 15 '14 at 11:53
  • I love the idea but, depending on the context, even the same person could be classified as different type of user, from project to project. Trying to identify some common behaviours is still a nice project though – Leths Oct 15 '14 at 12:54
  • Going through the user profiling techniques as part of the segmentation analysis is such an intriguing thought - I'll definitely do some more research when time permits! – Andreas Wolters Oct 15 '14 at 13:22

Market research people talk about customer/market segmentation analysis, which involves some of these types of segmentation (there are probably more):

  • Geographical: based on the physical location/region of the customer
  • Demographic: based on age, gender, racial, etc
  • Behavioural: types of consumer behaviour (but we can convert this to user mental models)
  • Cultural: important in the context of different language and nationalities
  • Needs: I think this can be linked to the Kano model of user needs (see related questions in UXSE)

Then you can create cross sections of these segments to come up with different types of users for different projects, and that should give you a common way to collate and build on the 'customer knowledgebase' that you are working towards.


Fuzzy logic

No system is perfect; you're cataloging things based on layers of interpretation. I'm in the midst of defining such a process for a new team now. Every time I do this, I find myself reconsidering the format. But it gets the job done.

Here's where I am at the moment. The specifics will change based on the application, but the fundamentals are there.

Basic "insight" data points

I start with defining the key pieces of data about any and every insight.

  • Source of contact (feedback link, cust serv call, survey, etc)
  • Product
  • Feature
  • User type / Persona
  • Insight type (praise, criticism, idea, random)
  • Attitude (pos, neg, neutral)
  • User journey stage (if you have it well defined)
  • Date/time stamp
  • Impacted KPIs (a small set of global factors)

Action items

I also like to map back to action items (link to your system of choice). Sometimes this means creating one. In the past, I have favored breaking up feedback that impacts multiple items. It's easier to sort the list, but you really have to tag each insight with an ID at that point so you can cross-reference the decomposed item.

  • Job/user story
  • Sprint backlog item

I dump all of this ... wait for it ... in a spreadsheet :( But it works. From there, I can generate various reports to share with the team. I almost never actually present the spreadsheet, but I've had people who like to look at it. It's just a tool for me to run pivots off of.

In raw form, I also have an Evernote notebook stuffed full of the most insightful stuff. I'll copy the note link over to my spreadsheet, presentations, or even job stories for reference.


Building up a database like this sounds pretty cool, and potentially very useful. The difficulty lies in understanding what the data will be used for. One way to start scoping this out would be to make a list of the top 20 questions you have about users at the beginning of any project. This will provide you a path to defining a clear list of dimensions and metrics and KPIs that you can use going forwards.

Once you have that list, you can start theorizing about what other data might be helpful to further understand users. For example, you might collect data about what login method a user has chosen. If they login with Twitter, you might want to know if they actively tweet (days since last tweet?), and try to use that data to inform your choices about exposing social sharing callouts in your product.

Be wary of overdoing it, though. You could find yourself scoping out a list of dimensions and metrics so deep and so wide that implementing it is impractical, or even if the implementation is practical, nobody will ever care about the more esoteric data being stored.


What I have done for cultivating a user research 'database' is actually several things:

  1. Personas
  2. User needs roadmapping - mapping all the themes I am hearing for new features along a timeline, with cross references to specific customers
  3. User maps - I take specific goals that customers are trying to achieve and segment maybe our top customers against that - exploring other dimensions at the same time. For example, to accomplish goal X, users can currently do options A, B,C or D. Mapping the top customers against this profile gives some interesting insights.
  4. I also maintain an Excel database of customer pains or needs that can be revisited

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.