In short: Feedier calculates several standard indicators from your questions and verbatims. This article gives the definition and formula for each one, so you know exactly what your numbers measure.
Overview
KPI | What it measures | Scale | Calculated from |
Satisfaction ratio | A feedback's overall satisfaction | 0 to 100% | Ratings, NPS, emojis and text sentiment |
NPS | Likelihood to recommend | −100 to +100 | NPS questions |
Sentiment | A verbatim's tone | 0 to 100 | Texts, analyzed by AI |
CES | The effort perceived by the customer | The question's scale | CES questions |
Business impact | A topic's priority | 0 to 10 | Volume, sentiment, NPS, CES, ratings |
NPS impact and effort score | The value and difficulty of an improvement area | 0 to 100 / 1 to 5 | Action plans |
Satisfaction ratio
It's calculated for each feedback, then averaged. It combines two families of elements, weighted equally:
quantitative elements: satisfaction scales and grids, emojis, NPS, converted to a percentage (7/10 = 70%);
qualitative elements: the sentiment of text responses.
Feedback element | Response | As % |
Rating | 7 / 10 | 70% |
NPS | 6 / 10 | 60% |
Satisfaction grid (average) | 4 / 5 | 80% |
Text 1 | Neutral | 50% |
Text 2 | Positive | 90% |
Calculation: quantitative average = (70 + 60 + 80) ÷ 3 = 70%; qualitative average = (50 + 90) ÷ 2 = 70%; satisfaction ratio = (70 + 70) ÷ 2 = 70%.
In the survey editor, the questions included in the calculation are marked with the % symbol. To exclude a text question from the calculation, turn off the option at the question's Advanced step.
NPS
NPS = % promoters − % detractors
Profile | Scores |
Promoters | 9 and 10 |
Passives | 7 and 8 |
Detractors | 0 to 6 |
Avoid having several NPS questions in the same survey: the average of the feedback's NPS scores is then used.
Sentiment
Each verbatim, and each passage linked to a topic, receives a score from 0 to 100.
Score | Sentiment | Typical emotions |
61 to 100 | Positive | Joy, trust, serenity, admiration |
41 to 60 | Neutral | No marked emotion |
0 to 40 | Negative | Anger, contempt, sadness, disgust |
At the topic or segment level, it's the average of the verbatims concerned.
Business impact
Sorting topics by volume means prioritizing what people talk about most, not what costs the most. Business impact scores each topic from 0 to 10: the higher it is, the higher the topic's priority.
Step 1: each signal is scored out of 100
Signal | Calculation | Base weight |
Sentiment |
| 45 |
NPS |
| 35 |
CES |
| 30 |
Ratings |
| 30 |
Volume | The topic's share of verbatims over the period | 25 |
All signals point the same way: a high value signals a problem. The 1.7 exponent gives much more weight to a topic that's very poorly received than to one that's merely lukewarm. A passive weighs about 3.3 times less than a detractor; a promoter doesn't count.
Step 2: only available signals count
A missing signal (no CES in the survey, for example) isn't included in the calculation.
The weights of the signals present are scaled back to 100%.
Volume is always included. Weights are fixed.
Step 3: the result is brought back to a score out of 10
Business impact = (Σ signal score × its weight) ÷ 10
Example
Topic "Delivery times": 60 verbatims out of 400, average sentiment of 35, 45% detractors and 25% passives. The survey has neither CES nor ratings.
Signal | Score /100 | Weight | Contribution |
Volume | 15.0 | 23.8% | 3.6 |
Sentiment | 48.1 | 42.9% | 20.6 |
NPS | 52.5 | 33.3% | 17.5 |
Total | 100% | 41.7 → 4.2 / 10 |
Key takeaway: the same topic can have a different impact from one survey to another, because the available signals change. Compare topics within the same survey first. There's no minimum number of verbatims: always read the score alongside the volume. A topic with no feedback over the period has no score.
Action plan scores
Score | Scale | What it indicates |
NPS impact | 0 to 100 | The potential NPS gain if the improvement area is addressed, based on the breakdown of its promoters, passives and detractors. |
Effort score | 1 to 5 | The complexity and resources required, estimated by AI. |
FAQ
My NPS differs from my former tool's. Why?
Check the period, the filters, the time zone, and whether the other tool counts partial responses. Feedier only uses NPS-type questions.
How do I create an indicator that isn't in this list?
With KPI formulas: response rate, detractor rate, CSAT, resolution time…
Last reviewed: November 11, 2026, version 4.23.5.
