What a survey reports
One number
A single rolled-up score: a 7.4, a 4-star, an NPS of 32. It smooths every disagreement into one tidy figure that flatters the dashboard and tells you nothing about who is actually angry.
On the card
Customer stories stay on the card.
Customers share stories with the work. Rank the backlog from that evidence, not a survey average.
signifier · engagement matrix
12 signifiers · same question · one “5” average
naïve average: 5.0two camps, not one
What the average hides
One mean. Two realities. The signifier keeps the distribution so the split is impossible to miss.
Why the average lies
A survey score hides the split.
Averages collapse two customer realities into a number nobody feels. ScrumDo keeps the stories on the card, so planning sees the camps that actually exist.
What the stories show
Two camps
The same stories, kept with the work, resolve into two clusters: a delighted group and a frustrated one. Almost nobody sits at the average the survey reported.
What becomes work
Source-linked patterns
Each camp becomes a named pattern attached to the stories and cards it touches, so planning can respond to the people who are actually there.
On the work
Stories stay attached to the card.
Distribution
See both camps
A survey score averages away the two camps inside it. Stories on the card keep the real split so you act on what customers actually feel.
Ungameable
The customer assigns the meaning
No obvious “good” answer to solicit, and no analyst recoding the answer after the fact. Customers keep more control over what their own story means.
It becomes work
Stories flow onto the card
Patterns travel into planning, cards, and report-back conversations, and on Business into WSJF, attached to the stories and interpretation that made them matter.
Proof
Trusted where customer signal has to be real.
Teams at regulated, customer-heavy organizations use ScrumDo to keep customer experience interpretable, source-linked, and close enough to the work to change what happens next.
Works with the agents you already run
ClaudeCodexGrokGitHub CopilotCursor
Stories on the card
Customers place their own experience next to the work.
See the split
See split customer realities an average hides.
Rank the work
Patterns shape the work without losing the stories behind them.
It doesn't stop there.
Bring your own agent
Make the agent accountable to the work
The same stories on the card can help your own Claude, Codex, or Copilot draft from a governed spec: no resold tokens, no lock-in, every run on the record.
See how →
Flow economics
Rank by what delay actually costs
On Business, customer patterns meet cost of delay and WSJF, so “which thing first?” becomes a number you can defend, not a loudest-voice argument.
See how →Customer stories
Rank the work from the stories you have.
Keep customer stories on the card. See the camps a score hides, and shape the backlog around the people who are actually there.