Customer feedback management
Forty-one customers had already told them the answer, in three different places.
Nobody had read them as a set. Watch 1,100 scattered support conversations, sales notes and interviews turn into one opportunity with 41 pieces of evidence behind it.
Feedback arrives in quantity. It gets read one item at a time.
Step 01
Collecting feedback was never the problem
FlowDesk sells shift scheduling to mid-size logistics operators. 400 customers, mostly Benelux.
They had 1,100 support conversations, notes from 60 sales calls, and 14 churn interviews someone had done in February and filed in a folder.
Every one of those was captured on purpose by somebody doing their job properly. All of it was there. It had simply never been in the same place at the same time, so nobody had ever read it as a single body of evidence.
A backlog you read one row at a time will always agree with whatever you already believe.
Step 02
Twelve people, one request, twelve wordings
Duplicate detection is what makes volume visible.
Nobody wrote "we need a second seat." They wrote about screenshots, texting the roster, view-only access, and the depot team.
Read one at a time, each is a small inconvenience worth a polite reply. Read together, they are twelve customers describing the same missing capability. A keyword search would have found none of them, because they share no keyword.
Step 03
Sort by what it's about
Where a request arrived is metadata. What it's about is the unit of work.
Sorting by source tells you which team is busiest. Sorting by theme tells you what your product is missing.
The three largest clusters in FlowDesk's 1,100 items came out at 41, 18 and 11. The largest drew on three sources that had never been in the same room: support tickets, a sales engineer's call notes, and the February interviews.
The second cluster, setup and import effort, had 18 items. That one mattered later. It was the evidence behind the solution their CEO wanted, and it was smaller than everyone assumed.
Step 04
All 41 came from accounts that had finished onboarding
This is where feedback stops being a wish list and becomes an argument.
FlowDesk's Q4 objective assumed onboarding was too hard. Cross-referencing the 41 items against the churn cohort put that in doubt: every one of them came from an account that had already been through it.
They had imported their rosters. They had published a first schedule. Then one person was alone in the product, and the thing they kept asking for was a way to bring somebody else in.
The most valuable thing feedback did that quarter was correct the reason for building, before anything got built.
Step 05
Promote it, and it stops being a line in a list
Some requests deserve a real look first.
The cluster became an opportunity — the second user never arrives — carrying all 41 items with it, linked in both directions.
That two-way link is the part that matters six months later. Open the opportunity and you can read the original words. Open any one of the 41 requests and you can see what happened to it. Nobody has to remember who said what in March.
From here it leaves this page. The opportunity gets scored against the objective, three solutions get proposed, and the cheapest assumption gets tested.
Tagging tells you a request exists. Citing it lets a decision point at the sentence that caused it.
Step 06 · 14 November
All 41 people who asked were told it existed
Because the release traced back through the opportunity to the original requests, the list of people to tell was already assembled. Nobody rebuilt it from memory.
Nine replied. Two became reference customers. Closing this half of the loop is the reason the next round of feedback arrives at all.
Ideas and pain points, side by side
Fair questions
What people ask about feedback management
- Where does customer feedback come from?
Support tools, chat, surveys, analytics, a native feedback portal, and bulk import for anything already sitting in a spreadsheet. In the demo, the three richest sources were support conversations, sales call notes, and a folder of interviews nobody had opened since February.
- How does duplicate detection work if people use different words?
It compares meaning rather than wording, then shows you the candidates with a similarity score and lets you decide. The twelve requests in the demo share no keyword at all.
- Does every request have to become a feature?
No. That's why a cluster gets promoted to an opportunity, where it still has to earn its place against an objective. Two of the three clusters in the demo never became anything.
- Can customers see what happened to their request?
Yes. The link runs both ways, so a shipped release knows which requests it answers and every requester can be told. That list assembles itself.
- We only get a handful of requests a month. Is this overkill?
The value comes from reading across sources rather than from volume. The reframe in the demo comes from 41 items, and it had been missed for months because those 41 lived in three different places.
- How long does it take to get anything out of it?
It depends on how fast you can connect what you already have. For a ten-person team that has typically taken about five days.
What have your customers already told you that nobody has read together?
Import what you already have. The first useful thing you find is usually a belief you'd been planning around.
FlowDesk is the Shorter Loop demo workspace. The clustering, the two-way links and the traced release is live in it — open it and check any number here.