There are eight types of customer feedback worth sorting by: feature requests, bug reports, usability friction, praise, churn reasons, pricing objections, support questions, and NPS or CSAT verbatims. Most lists stop at naming them. This one is a working taxonomy: for each type you get an example, the signal it carries, the team that owns it, where to route it, and the single metric to track. Sort by these eight and every message has somewhere to go.

Unless a quote links to a source, the examples are illustrative: we wrote them to show what each type looks like. For real examples from Amazon, Slack, Netflix and others, see 15 customer feedback examples.

This post is the hub of our customer feedback series:

The taxonomy at a glance

Type Signal it carries Owner Route it to Metric to track
Feature request Demand for something you don't have Product manager Feedback board Unique requesters per request
Bug report Something is broken Engineering lead Issue tracker (Jira, Linear) Time to fix
Usability friction It works, but people get stuck Product designer Design backlog, tagged by flow Repeat reports per flow
Praise What to protect and repeat Marketing, product Testimonial file, feature notes Mentions per feature
Churn reason Why revenue walked out Customer success Churn log, reviewed monthly Revenue lost per reason
Pricing objection Price or packaging mismatch Founder or pricing owner Pricing log Deals lost with price as the stated reason
Support question Docs or UI don't answer it Support lead Help center backlog Repeat questions per topic
NPS or CSAT verbatim The "why" behind a score Whoever owns the theme Tagged into the types above Share of comments per theme

Two things to notice. First, most types have a different owner. Feedback rots when it sits in a shared inbox that belongs to everyone. Second, every metric is a count or a duration you can pull in a minute. If tracking a type takes a dashboard project, you will stop tracking it.

1. Feature requests

Example: "Can we export the roadmap as a PDF for our quarterly board meeting?"

Signal: demand. A customer needs an outcome your product doesn't deliver yet. The words describe a solution; the job behind them (here, showing progress to people who never log in) is what you build for. The feature request glossary entry has the full definition.

Owner: the product manager for that area.

Route it to: one public or private feedback board, whatever channel it arrived through. Merge duplicates so the count is honest. FeatureOS feature request software does the collect, merge and notify steps.

Metric: unique requesters per request. Raw votes reward whoever shares the link most. Distinct accounts asking, ideally weighted by revenue, is closer to real demand. For how votes connect to a roadmap and a changelog, see feedback voting, roadmap and changelog in one tool.

2. Bug reports

Example: "When I filter the board by tag, posts from last week disappear. Chrome on macOS."

Signal: breakage. Something that used to work, or should work, doesn't.

Owner: the engineering lead or on-call engineer.

Route it to: your issue tracker, with steps and environment attached. Keep bugs off the feature request board. A crash should never compete for votes with a nice-to-have.

Metric: time from report to fix. Count customers affected as a secondary number so you fix the bug blocking 40 accounts before the one blocking 1.

3. Usability friction

Example: "I didn't realize I had to click Save before switching tabs. Lost my changes twice."

Signal: the feature exists but the flow fails people. It often arrives disguised as a "where do I find..." question or a complaint that the product is "confusing."

Owner: the product designer for that flow.

Route it to: a design backlog, tagged by the flow (onboarding, import, settings). One report is an anecdote. Three people tripping on the same step is a pattern.

Metric: repeat reports per flow. When the count for a flow drops after a redesign, the fix worked. If it doesn't drop, you fixed the help article and not the flow.

4. Praise

Example: "Merging duplicate posts used to take me an hour a week. Now it's a click."

Signal: what's working, in the customer's own words. Teams skip this type because nothing needs fixing. That's a mistake: praise tells you what not to break in the next redesign, and it gives marketing copy that sounds like a customer.

Owner: marketing for the quotes, product for the pattern.

Route it to: a testimonial file (with permission to quote) and a note on the feature it's about.

Metric: mentions per feature. The features people praise unprompted are your real positioning, whatever the homepage says.

5. Churn reasons

Example: Cancel form: "We moved to a tool that integrates with our CRM."

Signal: lost revenue, with a stated cause. The stated cause isn't always the real one, so a short follow-up call is worth more than the form.

Owner: customer success.

Route it to: a churn log, reviewed monthly next to the feature request board. When a churn reason matches an open request, link them: that request now has revenue attached. Watch your overall churn rate alongside it.

Metric: revenue lost per reason, not account count. Ten small accounts leaving over a missing export may matter less than one large account leaving over SSO.

6. Pricing objections

Example: "Most of our team would only log in once a month. Paying a full seat price for each of them is hard to justify."

Signal: a mismatch between what you charge and what the buyer thinks it's worth, or how they want to buy it. These come from sales calls, trial-end emails and cancel forms.

Owner: whoever sets pricing. In a small company that's the founder.

Route it to: a pricing log, separate from churn reasons, with the plan, team size and competitor named if there was one. Kept apart, objections show whether the problem is the price, the packaging (seats, tiers, limits) or the comparison.

Metric: deals lost with price as the stated reason, per quarter. One objection is a negotiation. The same objection from the same segment every month is a packaging decision.

7. Support questions

Example: "Is there a way to bulk-archive old posts? Doing it one by one is taking forever."

Signal: the product or the docs failed to answer something. Many support questions are hidden feature requests or usability friction; this one is both.

Owner: the support lead.

Route it to: the help center backlog when the answer exists but can't be found. When the answer is "no, you can't," log a feature request against the customer's name. In FeatureOS, agents can create or upvote a post from Intercom, Zendesk or Salesforce without leaving the ticket.

Metric: repeat questions per topic. A question asked ten times a month is a docs gap or a UI gap. Fix the one it is.

8. NPS and CSAT verbatims

Example: NPS score 4: "Good product, but the mobile app is too slow to use on site visits."

Signal: the reason behind a score. NPS asks how likely someone is to recommend you on a 0-10 scale (promoters 9-10, detractors 0-6; our NPS calculator does the math). CSAT asks how satisfied they were with one interaction, usually 1-5. The score is a trend line. The comment is the work.

Owner: nobody owns verbatims as a type. They get split up.

Route it to: the other seven types. The example above is usability friction or a bug (mobile performance), so it goes where those go. Reply to detractors within days, then tag the comment by type and theme. More on CSAT in our post on satisfaction survey forms.

Metric: share of comments per theme. If "mobile speed" goes from a third of detractor comments to a tenth after a release, the release worked, whatever the headline score did.

Route it in 60 seconds

When a new piece of feedback lands, run down this list and stop at the first yes.

  1. Is something broken? Bug report. Send it to the issue tracker with steps.
  2. Is the customer leaving or has left? Churn reason. Log it with the account's revenue.
  3. Is the complaint about price, plan or seats? Pricing objection. Log it with plan and team size.
  4. Does it ask for something the product can't do? Feature request. Post it on the board or upvote the existing post.
  5. Can they do it, but couldn't figure out how? If the answer is in the docs and they missed it, support question. If the flow itself tripped them up, usability friction.
  6. Is it a thank-you or a compliment? Praise. Save the quote, ask permission to use it.
  7. Is it a comment attached to a score? NPS or CSAT verbatim. Reply first, then run the comment through steps 1-6.

Two rules keep this honest. One piece of feedback can be two types (a churn reason that names a missing feature): log it in both places and link them. And the person routing it tells the customer where it went. That reply is the first step in closing the feedback loop.

What we see on our own board

We run FeatureOS on FeatureOS, at feedback.featureos.app. The public side has three boards: Feature Requests, Integrations, and Support Desk. That split is the taxonomy above in practice: feature requests and integration gaps get voted on, while support questions get their own board, so how-to questions don't sit in the vote rankings next to product ideas.

One post shows the whole path. In April 2022 a customer asked for custom OG images for each post, so shared links would preview better on social media. Two other posts asked for the same thing in different words; both were merged into it. It reached 48 upvotes. We shipped it, posted the changelog Dynamic OG image for Posts on April 28, 2024, and marked the post Completed. Without the merge, that demand would have looked like three small requests instead of one large one.

Put the taxonomy to work

Pick the three types you lose most often (for most teams: churn reasons, pricing objections and praise, because they arrive in places product never looks) and give each an owner this week. Then:

  1. Capture every type in one system, whatever channel it came from (how to collect customer feedback).
  2. Tag each item by type and theme, and merge duplicates (customer feedback analysis).
  3. Decide what to build, fix or decline using votes, revenue and effort. The feature prioritization calculator helps.
  4. Ship and announce on a public roadmap and in your release notes.
  5. Notify everyone who asked.

FeatureOS covers those steps in one place: feedback boards, AI merge suggestions, roadmaps, changelogs and voter notifications. See customer feedback software for the overview. Plans start at $60/month for 5 seats, and you pay only for team seats; see pricing for details and the 14-day free trial.