Customer Satisfaction Dashboard
Three Unhappy Clients. One 70% Disaster.
How a Consulting Firm Rebuilt Its Customer Satisfaction Dashboard Around Sample Size
A satisfaction percentage feels solid. It has a decimal, it moves up and down, and it gets reported like any other metric. What the percentage hides is how many people it came from. A number from ten responses and a number from ten thousand look identical on a dashboard. They behave completely differently.
A consulting firm was measuring client satisfaction the way most firms do: a 1 to 5 rating after each engagement, rolled up into a percentage. The questions were fine. The issue was the count. Their engagements were large and their client list was short, so each quarter produced only a handful of ratings. A percentage came out of that handful, but it rested on so few people that a single client could swing it.
One client was worth ten points
With ten responses, each person moves the score by ten points. Three unhappy clients out of ten reads as 70%, and it looks like a real drop. The same three clients out of a thousand wouldn’t move the number at all. At that sample size, the score tracks whichever three people happened to answer more than it tracks the quality of the work.
That’s why the number swung so hard from quarter to quarter. A couple of rough engagements, or even a couple of clients in a bad mood, could send it lurching while the underlying quality of the work held steady. The percentage looked precise and behaved like noise.
Never a score without its sample size
The fix was structural. We built a dashboard where every satisfaction figure sits next to the number of responses behind it. Nobody sees a 70% on its own. They see a 70% that came from ten people, right beside a 92% that came from four hundred, and the difference between the two is impossible to miss.
Putting the denominator next to the percentage changes how the number gets read. A score from ten responses stops looking like a verdict and starts looking like what it is: a reading with a wide margin, one that a single response can push around.
They stopped reacting to individuals
With the sample size visible beside every score, leadership could tell a real decline from three people having a rough month. A dip built on ten responses got read as noise until more data came in. A dip built on hundreds got taken seriously right away.
That freed up attention for the drops that were signal. The firm stopped chasing every quarterly wobble and started spending its energy on the movements that meant something, which were the ones backed by enough responses to trust.
The Lesson
A percentage without a denominator is a rumor. At ten responses, one person is worth ten points, so the number can look precise while swinging on individuals. Small samples give you a score that moves on who happened to answer, and reading it as a stable measure of quality leads you to react to noise.
A short client list produces few responses no matter what you do, so the sample would always be small. The dashboard handles that by showing the count every time. A number gets the trust it has earned and no more, and a 70% from ten people never gets to look like a 70% from a thousand.
Show the Denominator Beside the Score
At ten responses, one client is worth ten points, and a satisfaction percentage can swing on whoever happened to answer that quarter. Reported without its response count, that number can look like a verdict when it’s really noise. We build the score and its sample size into the same view, so a number gets read for exactly as much as it’s worth. And if your response volume is already high enough to trust the number as it stands, we’ll tell you that. Reach out to us and book a free Data Health Check.
