Passive

Sorting survey answers into good and bad makes the middle disappear. Count 7s and 8s as satisfied and the score comes out kinder than the situation is; count them as unhappy and customers with nothing wrong get logged as problems. Either placement makes the number describe something that isn't happening.

The size of the effect is easy to show. In a survey of 200 responses with 90 people at 9 or 10, 60 people at 7 or 8, and 50 people between 0 and 6, putting those 60 on the satisfied side gives a score of 50, and leaving them out of the calculation gives 20. Same responses, thirty points apart. So this group got its own name and was taken out of the arithmetic instead of being folded into either side.

Answers of 7 and 8 land here. In the calculation they appear in the denominator and never in the numerator. In the example above the score is the promoter share of 45 percent minus the detractor share of 25 percent, which is 20; those 60 people sit in neither figure while still enlarging the 200 that both are divided by. So when this group grows, the score gets pushed toward zero on its own.

They are held separate because their behavior differs from both neighbors. They're using the product and filing no tickets, but they move when something comparable shows up slightly cheaper. At Pagedeck, where teams answering 9 or 10 renewed at 88 percent the following year and teams answering 0 to 6 renewed at 41 percent, teams answering 7 or 8 renewed at 68 percent — in between. Fold them into the satisfied side and that twenty-point gap stops being visible.

In practice their value shows up in the free-text box. A detractor comment tells you what is broken; a comment from this group tells you the one condition under which they would recommend it. They are also usually the largest of the three, so movement from here into promoter shifts the score more than anything else.

"Passives are 30 percent. Let's read their comments first."

This is where the most room to lift the score sits.

"Don't roll 7s and 8s in with the satisfied group. Their renewal rate is twenty points lower."

Giving the numeric reason for keeping them out of the calculation.

"The score didn't drop — the passive share grew."

Explaining a falling score in a quarter when detractors didn't increase.

Pagedeck hit exactly that last quarter. The score went from 24 to 18 while the count of people answering 0 to 6 stayed flat; 9s and 10s had shifted down into 7s and 8s. Mia read those respondents' comments and found the same thing said twenty-six times: search was slow. It was never bad enough to file a ticket about, so the support log held not one mention of it.