Reading customer feedback data to surface what matters and what to do next.
Your boss is unhappy about last month's NPS drop and wants reasons. Before you start hunting for causes, check whether the drop is within margin of error, i.e. did it actually change! Here's how to calculate NPS margin of error (two ways), the Chi-squared test, and which to use when. Or skip ahead and download the free Excel spreadsheet.
When someone gives you a zero on NPS, they aren't just rating you, they're sending a message. The data backs it up: response rate and comment rate curves spike at zero across multiple companies in a remarkably consistent shape. Here's the analysis, why it matters for service recovery design, and the case for treating zeros as a separate workflow.
An aggregate NPS number tells you very little about which customers are loving you and which are quietly leaving. Segmentation does the heavy lifting, demographic for B2C, firmographic for B2B, behavioural for both. Here's how to slice NPS data effectively, with the rules that stop segmentation from devolving into noise.
Where should you spend the next CX dollar — service responsiveness, a new feature, queue times, the website? The Kano model gives you a way to triage by customer impact. Combined with Net Promoter Score data, you can identify which attributes are hygiene factors, which delight, and which to leave alone. Here's how to run the combined analysis.
Your dashboard says NPS is up 15 points week-on-week. Email's already drafted. Then you stop, was that real, or just sampling noise? Take credit for the rise this week and you'll be taking blame for the drop next week. Here's how to tell whether NPS has really changed, with a downloadable spreadsheet to check for yourself.
Three minutes after your first NPS survey closes, management will want a report. The question is what to put in it beyond the score chart. Here are the nine elements that every NPS report needs, plus what to deliberately exclude, so the report drives action instead of arguments about sample size and "that score can't be right".