KjerrObservations
The distinct problems and requests. Ideas from different people that ask for the same thing are pooled into one observation, so forty reports of one bug are one row, not forty.
A feedback analysis tool
that shows its evidence.
Kjerr's Insights tab is a feedback analysis tool built into the feedback board. With one switch, AI reads every idea and the comments on it, pools what people ask for into observations, keeps what they praise as strengths, files both under themes, and ranks each by how many distinct people are behind it. Every number opens the ideas and the quotes it was counted from.
What the feedback analysis finds
Three lists, each a different question about the same feedback. Together they say what people want, what they value and which parts of the product the feedback is about.
Strengths
What people explicitly value, in their own words. Useful for knowing what to protect, and for quoting them.
Themes
The areas of your product the observations fall under, such as Billing or Search. Where your topics fit, a theme takes a topic's name.
Ranked by reach, not by noise
Each observation and strength is ranked by its reach: the distinct people behind it, counting the authors of its ideas, everyone who voted for them, and the people quoted from the comments. Someone who appears in five ideas counts once, so the ranking reflects demand, not volume. Fewer than two people is Emerging; two or more is Active.
- RequestDark mode for the dashboardActive48 people
- ProblemCSV export times out on large accountsActive31 people
- RequestA Slack message when a status changesActive19 people
- ProblemSearch ignores accented charactersEmerging1 person
Each row opens the ideas it was counted from, with the sentence every point came from.
How to turn it on
- Open Insights in the top menu. It sits beside Reports. Viewing it needs the reports permission; switching it on needs a workspace admin.
- Click Turn on Insights. A bar shows the ideas being read. A board of a few hundred ideas takes a few minutes, and you can leave the page.
- New feedback follows. From then on every new idea, and every comment a visitor leaves on one, is read as it arrives. An edited idea is read again.
- Check the evidence. Click a row and the right-hand pane lists the ideas behind it, most voted first, with each point found and the sentence it came from.
- Resolve or dismiss. Mark an observation resolved when the problem is handled, or dismiss what is not actionable. Reopen puts it back under what reach says.
From a count to the words behind it
A number you cannot open is a number you have to trust. Here every one of them is a link.
- Open in inbox takes you from any idea in the evidence to the idea itself, in the feedback tracker.
- An idea in the inbox shows the observations and strengths it feeds, under its votes.
- Filter by status, and by where the ideas came from: Portal, Added by team, Imported, API.
- Sort by most people, newest or most ideas. Export CSV downloads the tab as it is filtered.
- On a multilingual board, a comment the board translated is quoted in the board language.
- Themes prefer your topic names, so the groups line up with the customer feedback analysis you already do by topic.
What this feedback analysis tool leaves out
Insights reads a feedback board. It is not a survey platform or a text-mining suite, and these are deliberately not included:
- Sentiment scores. An observation is a problem or a request, with the quotes; nobody is given a mood.
- Coding of survey answers, NPS verbatims or support tickets. Only the ideas and comments on the board are read.
- Automatic tagging or topic assignment on the board. Themes live in Insights; your topics stay yours.
- Generated replies or an AI agent that answers customers.
- A priority score. Insights ranks by reach only; RICE scoring is a separate job on the internal roadmap, and who asked is on Users & Companies.
- Mining of calls, chats or tickets from other tools.
What is sent out, and what stays
Idea titles, idea text and the public comments on them are sent to OpenAI to be read and sorted. Names, email addresses, votes and internal notes stay in Kjerr, and nothing on the board changes: Insights only reads. Turn the switch off and what was found stays on the page; nothing new is read until you turn it on again.
This is one of the four jobs AI does in Kjerr, beside suggesting duplicates for a teammate to approve, translating the board and proposing roadmap scores a teammate applies. It does not tag ideas, write replies or decide anything. Your own assistant can read the same board through the MCP server.
Who it is for
Product teams at SaaS companies of any size, from startups to enterprise, that want a public feedback board, a team-only roadmap scored with RICE, and reliable follow-up emails without a help-desk suite or a per-seat bill. Insights earns its place once a board holds more ideas than one person reads in a sitting, and the question changes from “what came in?” to “what does it add up to?” The follow-up, telling people what you did, is how you close the loop.
Questions, answered plainly.
What is a feedback analysis tool?
A feedback analysis tool reads a body of customer feedback and groups it, so a team sees the problems and requests that recur instead of a list of individual messages. In Kjerr the tool is the Insights tab: it reads the ideas and comments on a feedback board and sorts them into observations, strengths and themes, each ranked by the number of people behind it.
How is this different from the Reports page?
Reports counts: feedback per week, by status, topic and source, time to ship, what is awaiting a response. Insights reads: what the ideas say, grouped by meaning. Reports answers how much and how fast; Insights answers what, and who.
Does the AI change anything on the board?
No. Insights only reads. It does not tag ideas, merge them, set statuses or write replies. A teammate can mark an observation resolved or dismiss it, and that changes nothing on the board either.
How is reach counted?
Reach is the number of distinct people behind an observation or strength: the authors of its ideas, everyone who voted for them, and the people quoted from the comments. A person who appears in several ideas is counted once. Imported votes without a named voter count once per idea. Companies are counted the same way when people belong to one.
What does Insights cost?
Kjerr is free during early access, with every feature included and no credit card. When paid plans arrive they will be priced per workspace, never per admin seat and never per tracked user, so the number of people who post, vote or comment never changes the bill. Insights is included; there are no credits to buy.
What is sent to the AI, and what is not?
Idea titles, idea text and the public comments on them are sent to OpenAI to be read and sorted. Names, email addresses, votes and internal notes stay in Kjerr. A board that leaves the switch off sends nothing.
Related pages
Customer feedback analysisThe five steps, and where Insights fits.
Feedback trackerThe inbox each observation links back to.
Feature request toolWhere the ideas, votes and comments come from.
Multilingual feedbackTranslated comments are quoted in the board language.
Give your users an answer.
Kjerr is free during early access, with every feature included and no credit card. When paid plans arrive they will be priced per workspace, never per admin seat and never per tracked user, so the number of people who post, vote or comment never changes the bill.