How to build a customer feedback taxonomy
Are you recording customer feedback data across various data sources (customer meetings, social media, support, reviews) but unsure how to bring it all together? This article shares my experience with using a taxonomy to categorize and unify all customer feedback as a “360” source of truth, so the data can be used for business decisions. A theoretical UI for a simple internal tool you could build is available at the bottom of the post. I’ve used a similar approach at a certain tech giant that starts with a G, so let me know if you have any questions about getting started with managing your user feedback or on how to bring in data sources like user research interviews, app reviews, etc.
How to categorize your feedback
The system described here places metadata on tags and on incoming feedback. The incoming feedback has information like the customer reporting the feedback, the device, version of the software, and the date. The feedback is then tagged with feedback tags, which contain a description of a common issue and metadata about that issue, such as the feature area of the product it is related to, sub-features, and whether it is a bug, feature request, or some other type. Feedbacks coming in tend to have lots of tags, and tags belong to a lot of feedbacks. The customer is tracked if you want to engage with them for resolution or research.
Why should I analyze my user feedback?
Your users know what they want to pay you for, so you should listen to them. If they say there is too much product debt, address the bug backlog. If they want a shiny new feature, assess the business value. If they keep clicking on a button because it’s in a bad place, fix your UI.
By spending a few minutes or bringing on someone junior to tag your feedback, you’ll have a large amount of customer feedback that you can use to back business decisions, instead of anecdotes or a nag from your most vocal customers. If you need to say no to a customer, you have data to back up the decision.
Intuition is cool, but it’s not the best tool for product decisions.
How do I get my customers to leave feedback?
Report an Issue/ Leave Feedback Popup
There are many ways to get customer feedback. The most direct way is to leave a “Report an issue” or “Leave feedback” option that can be accessed from any screen in your app so customers get into the habit of leaving feedback. Here’s a tool for that: https://canny.io/features/collect-feedback
You can also build in an automated option via support to leave feedback and have those reports dump directly to your team instead of the support team. (Trust me, after working in a high volume support center I can assure you that they don’t want to filter that for you.) If you use something like Zendesk, you can add forms for the feedback tickets so customers can leave screenshots.
Ask! When you release new features, ask for feedback. When you’re doing research, ask for feedback. You’ll likely get a mix of praise and negative reactions, but the negative comments should cut into what isn’t working.
If you’re not already sending out surveys, you’re missing out on critical user feedback. Tack on a question like “If you had a magic wand, what would you add to our product” and you’ll get a wishlist from your customers.
Engage with other teams
Have teams submit feedback from customer meetings, things they find on the internet, reviews on review sites, commentary from your user forum, and more. If it’s feedback from a customer, you want to be counting it.
This is a lot of work, how do I handle it?
Instead of dumping this on your PM, you can hire someone junior to do all of the tagging and to present the findings at team all hands meetings or to the appropriate stakeholder. This resource can be shared across teams. Overall, they’ll be monitoring incoming customer feedback, and tagging it within an existing taxonomy. More info below on how to write that taxonomy. This person would need high level awareness of your product, but wouldn’t need to be a product expert.
In terms of using the data, this data can be used as the voice of the customer in a very strong way. Some examples on how to share the data include having a weekly theme that you dive into a high ranking feature request and read out some of the customer feedback and dig into their use cases, or seeing on a weekly basis what are the highest ranking items that are of a specific type, like usability issues or bugs. Basically, your feedback tagger would live and breathe this customer data and would pull metrics and support the PM team and help communicate the voice of the user to your engineering team and other stakeholders who need to understand the customer in this way.
What tool do I use?
Don’t use Google Sheets. Buy a product or make something in house. I’ve only used in-house solutions, so I’ve provided mocks below on what things I think are most critical for this. You might need to add some additional information to your database. This is a very simple tool unless you start adding in integrations with SFDC, Jira, and other tools for things like sentiment scoring or followup engagement, so you can knock it out quickly if you can grab a couple of engineers for a few days of planning and execution. Other than that, you can look into simpler options, like Canny, or more complex options, like what you see here, but you might spend more time getting disappointed in sales calls than it actually takes to build a super simple tagging tool. https://www.techradar.com/best/best-customer-feedback-tools#8-canny. Generally the ones I’ve seen lack in having multiple tagging abilities or in having the ability to add a type on the tags or a subcategory.
With SAAS product feedback, generally, we can categorize most feedbacks into bugs, usability issues, feature requests, knowledge captures, education requests, or marketing material requests. These TYPES are distinct from a product feature TAG, which should include a product feature and relevant sub-features.
Each feedback we receive will contain a bunch of ‘sub-feedbacks’, hence the need to tag on the feedback and store the tag’s information in a different table.
Let’s repeat that: the categorization and details for the tag belong on the tag rather than on the feedback. Imagine this coming in from a customer:
OMG I have had it today with this frigging CLEAR Button!!!!!!!!!!!!!!!!!!! I keep clicking on it while I’m trying to actually undo my last action and then I can’t even undo clearing it and I lose my data 😭😭😭😭😭😭. Can you PLEASE FIX IT ASAP OR I”M GOING TO LOSE MY MIND. Also, that new blablabla feature is super dope! Thanks for releasing that! I Hope you’ll add in blablabla soon! And fix this clear button plzzzzz — Annoyed but loyal customer
For this feedback, you’d need to add in a few tags:
In general, once you’ve received a feedback and have a tool in place and have created your taxonomy, tagging and categorizing is pretty straight-forward.
- Customer submits feedback
- Agent divides the feedback snippet into sub-feedbacks and tags for each feedback item
Example (hypothetical product is a drag and drop web builder with a text widget)
Hey, I really like the editor, but I can’t figure out how to change the line height in my text editor. I’m trying to make it be twice as big as the text size. Also, I wish I could format bullet points in the text editor because I don’t know CSS. Can you please help me make it have purple star bullets?
- Text editor > Line height Type: Usability issue “Line height feature in text editor is unknown”
- Text editor > Line Height Type: Feature Request “Custom line-height”
- Text editor > Bullet points Type: Feature Request “Modify bullet point design in TE”
- Developer Mode > CSS Type: Education “CSS training/ certification track”
- Text editor > Bullet points Type: How to “How to customize bullet point settings with CSS”
Intake of feedback from multiple tools:
- Site/in-editor feedback popup
- Manual or CSV upload
- Reviews from review platforms
- Social media posts
- Requester Name
- Request Email
- Way that it was submitted
- View/count tags by category, subcategory, date, type, etc
- View reports or exportable and run SQL reports
- Predictive tagging
- User sentiment
Category management space
Each post — Can be tagged with multiple tags, status, P0-P4
Tags — can be synced to JIRA
Cases — can be listed in JIRA linked tag
Tag — View all associated cases if viewing a specific tag
- View counts, show snapshot of the feedback pieces that are tagged
Other features to consider:
Integrate with SFDC to sync opportunities and dollar value, Jira to push cases to outages, bug reports, usability request, backlog tickets, etc., or other BI tools for high level visibility to other teams.
Here are some demo mocks I threw together: