How to Analyze NPS Survey Comments
min read - Updated on Nov 15, 2026To analyze NPS survey comments, keep the 0–10 score as a column next to each comment, create a Report from the file, compare the comments of promoters with those of detractors, and use Signals to find the topics that each group mentions more. Kimola groups every comment under the themes it mentions, labels its sentiment and answers research questions about it with Dimensions, so you can see what customers who score you 9 or 10 praise and what customers who score you 0 to 6 complain about. The score says how many customers are promoters or detractors, and the comments say why.
This guide takes an NPS survey export from the file to the topics that drive promoters and detractors. It works with an export from any survey tool in .xlsx or .csv format, and with a Google Sheet.
The Goal
An NPS survey asks customers how likely they are to recommend you, on a scale from 0 to 10, and usually adds an open question that asks for the reason for the score. Customers who answer 9 or 10 are promoters, those who answer 7 or 8 are passives, and those who answer 0 to 6 are detractors. The NPS is the share of promoters minus the share of detractors. That number shows whether loyalty is rising or falling. The reasons are in the comments, and a survey can collect thousands of them.
Customer experience, product and research teams analyze the comments after every survey wave. They ask what detractors talk about that promoters rarely mention, what promoters value, and which topics to fix first. This guide answers those questions with the Compare, Tables and Signals pages of a Report.
The survey NPS comes from the 0–10 scores, so it stays a column in your file. In a Space, Kimola calculates an NPS from the text of the feedback instead: the share of positive feedback minus the share of negative feedback, as described in How NPS Is Calculated in Spaces. Spaces are available in the Enterprise package. The sentiment of a comment and its score can also differ, because a customer can give a 6 and write a friendly comment.
The survey NPS comes from your score column. The NPS that Kimola shows in Spaces comes from the sentiment of the feedback text.
What You Need
You need a file with one row for each respondent, or a Google Sheet with the same layout. The comment of each respondent is in one column and the 0–10 score is in another. The date of each response is useful too. Columns such as the customer segment, plan, product, region or survey wave are worth keeping, because they become filters and groups in the Report.
You also need a Kimola account. Reports are available in every package. A Google Sheet needs the Google Sheets integration, which is available in the paid packages. Only rows with a comment are analyzed and counted in queries, and the rule is queries = rows × models × Dimensions. See kimola.com/pricing for the packages.
Step 1: Prepare the File with the Score Column
Export the survey results from your survey tool as an .xlsx or .csv file. Kimola accepts .xls, .xlsx, .csv and .tsv files, one file at a time, up to 30 MB. The first row holds the column names, and every row below it is one respondent. When your survey tool exports the comments and the scores in separate files, join them in a spreadsheet first, so that each comment sits in the same row as its score.
Rows with an empty comment can stay in the file. Only rows with text are analyzed and counted in queries.
Keep the 0–10 score as its own column. Kimola doesn't analyze it, so it stays in the Report as a column of your data that you can filter, compare and measure. The comment goes in the Text Column, which Kimola analyzes. If the file has the date of each response, keep it as well. A Date Column lets the Report show how topics change over time.
Add one more column with the NPS group of each score: Promoter for 9 and 10, Passive for 7 and 8, and Detractor for 0 to 6. In a spreadsheet, a formula such as =IF(B2>=9,"Promoter",IF(B2>=7,"Passive","Detractor")) fills the column when the score is in column B. Three groups are easier to compare than 11 separate scores, and a comparison in Kimola holds at most 5 parts.
[Screenshot: An NPS survey file with the comment, score, NPS group, plan, region and date columns]
Add a column with the NPS group of each score, such as Promoter, Passive and Detractor, so that Kimola has three groups to compare.
A mobile network operator exports its quarterly NPS survey with the columns Customer Comment, Score, NPS Group, Plan, Region and Answered On. Customer Comment is the Text Column and Answered On is the Date Column. Score, NPS Group, Plan and Region stay in the file as columns, so the team can compare promoters with detractors and read each plan or region on its own.
Step 2: Create the Report
Open Reports in the left menu and drop the file onto the upload area. For a Google Sheet, click Google Sheets under Integrations and pick the spreadsheet. Both ways open the same steps. Create a Report by Uploading a Custom Dataset and Connect Google Sheets to Kimola describe every step. This section lists the choices that matter for NPS comments.
In Sheets & Columns, choose the comment column as the Text Column and the response date as the Date Column. The score and the group column need no choice here. They stay in the Report as columns of your data.
[Screenshot: The Sheets & Columns step with the comment column chosen as the Text Column and the response date as the Date Column]
In Classifiers, choose the Automatic Classifier, which discovers the themes from your comments, and the Sentiment Classifier, which labels the sentiment of each comment. A content classifier and the Sentiment Classifier together also give the Report its best and worst performing themes.
[Screenshot: The Classifiers step with the Automatic Classifier and the Sentiment Classifier chosen]
In Structure, use Enable Multi-label Classification. NPS comments often mention several aspects at once, such as an app that is easy to use and a support team that took a week to answer, and a multi-label Report gives such a comment a label for each aspect. See Create a Report Using Multi-Label Classification.
In Dimensions, choose the questions to ask of every comment. Pain Points asks "What goes wrong or creates difficulty?" and shows why detractors score low. Motivations asks "Why is this chosen or preferred?" and shows why promoters recommend you. Unmet Needs asks "What is missing or desired?" and shows what customers still want. See Add Dimensions to a Report for every Dimension.
Each Dimension adds to the queries a Report uses: queries = rows × models × Dimensions. Choose the Dimensions you need.
In Review, give the Report a Report Title, such as the name of the survey and its wave, and choose the Report Language. Your comments can be in any of the 30+ languages Kimola analyzes, whatever the Report Language. See Language Options in Reports. Check Required Query and click Create Report. When the analysis is complete, Kimola shows Here we go!. Click View Report to open the Report on Overview, which shows the number of records analyzed, the sentiment, the best and worst performing themes and your Dimensions.
Step 3: Compare Promoters and Detractors
Compare sets groups of a Report's feedback side by side with the same analysis for each group. It reads the analysis the Report already has, so it uses no queries. Open Compare in the menu of the Report and click the group column you added in step 1, such as NPS Group, in the Compare by Columns bar. A list of its values opens. Choose Detractor first, then Promoter, and click Compare. Choose Passive as well if you want a third part. See Compare in Reports for every part of the page.
The first part you choose stays on the left, and its themes set the lines of every classifier list, so choosing Detractor first makes the table read from the detractors' side. Every other part lists the same themes on the same lines, with 0 comments where it has none.
[Screenshot: The comparison of the Detractor and Promoter parts with the Volume, Sentiment and Classifications rows]
Read the table from top to bottom. Volume shows the number of comments in each part and its share of the Report. It counts the comments Kimola analyzed, so respondents who left no comment aren't in it, and its shares can differ from the shares behind your survey NPS. Sentiment shows the Positive, Negative and None comments of each part. Mostly negative comments among detractors and mostly positive comments among promoters is the usual pattern, and a part that departs from it is worth reading.
Open the content classifier under Classifications to list its themes. Each theme shows its number of comments, its share of the part and the sentiment most of its comments carry, such as "97% Negative". Read the detractor column first for the themes detractors raise most, then read the same lines in the promoter column. Dimensions shows the answers of each Dimension for every part, with their numbers and shares: the Pain Points of detractors, the Motivations of promoters and the Unmet Needs of each group. Popular Terms lists the words used most often in each part.
A theme that promoters mention and detractors don't is listed under Mismatches at the end of the promoter part's list, with its number of comments, its share and its sentiment. Mismatches are often the most telling result of a comparison, because they name what one group talks about and the other group never does.
Choose the group you care about most first. Its themes set the lines of the list, and the extra themes of the other parts appear under Mismatches.
In the operator's Report, billing errors is a theme in 38% of the Detractor comments and in 3% of the Promoter comments, and most of its comments are negative. Network coverage and shop staff lead the Promoter part. The Pain Points of detractors include unexplained charges and long waits for support, and the Motivations of promoters include reliable coverage while traveling.
Compare shows the groups side by side. To see how the score itself moves by theme, open Tables in the menu of the Report, drag the content classifier into the table and measure the score column with Average. A column that holds numbers can be measured instead of used for grouping, and each theme then shows the average score of its comments, with the average of all comments beside it. The themes with the lowest averages are the ones that go with low scores. See Tables in Reports.
Step 4: Find Drivers with Signals
Compare shows how the groups differ. Signals are the statistically tested findings behind those differences: the topics that customers in one group mention more often, or more negatively, than the others. Kimola generates a Report's signals after its analysis completes, so the page is ready when you open Signals in the menu of the Report. See Signals in Reports.
The Differences tab lists the topics that stand apart in one group. The group is a value of one of your columns, so the group column from step 1 is what lets Kimola test promoters against detractors. Click All columns and choose that column to see only the signals about the NPS groups. A Difference can point up or down, so a topic that detractors rarely mention is listed too.
[Screenshot: A Differences signal comparing one region with the others, with the Region filter selected and “Why this is shown” expanded]
Every card sums up one finding. The two percentages give the measure in the group and in what it is compared with, such as the share of Detractor comments that mention a topic and the share of the other comments that do. Mentions says how much a topic is mentioned in the group, and Negativity says how negatively. Read the quotes on the card, then open Why this is shown to see what the group is Compared with:, the numbers behind the finding and its consistency level, which tells how well the finding holds when Kimola splits the comments into comparable groups.
Click Open in Tables to read the comments behind a finding and cross them with other columns, such as the plan or the region. Give each card a verdict, Known, New & important, Unimportant or Artifact, so your team can tell which findings it already knew and which ones need work. See Review Signals.
In the operator's Report, a Detractor card shows billing errors in 38% of the Detractor comments and in 8% of the other comments, with Mentions as its measure. A Promoter card shows shop staff in 22% of the Promoter comments and in 6% of the others. Open in Tables on the first card lists the detractor comments about billing, and crossing them with the Plan column shows that most of them come from one plan.
If your file has a Date Column, the Trends tab lists the topics whose mentions changed in the latest periods compared with the periods before, which shows what is rising or falling between survey waves. A signal needs enough comments in a group, and groups to compare. When the page says "No signal has enough evidence in this report", more comments, or more descriptive columns such as plan, product or region, give the tests more to work with.
More comments, or more descriptive columns such as plan, product or region, give the signal tests more to work with.
What You Get
The analysis gives you five results in one Report. Each one answers a part of the question of why customers score you as they do.
Themes by NPS group show the themes of promoters, passives and detractors with their number of comments, share and sentiment, and the themes that only one group mentions.
Dimension answers by group show the Pain Points, Motivations and Unmet Needs behind the scores, each with its number of comments and share.
Score by theme shows the average score of the comments in every theme, so you can see which topics go with low scores.
Drivers are the Signals that name the topics detractors or promoters mention more, or more negatively, than the others, with quotes, numbers and the comments behind them.
A result you can share comes from exporting the Report to Excel, CSV, PDF or PowerPoint, or receiving it by email.
Next Steps
To turn the drivers into work, create an Action Plan from the Report. Choose the side Our own offering and the goal Customer experience, and Kimola writes up to 12 actions from the evidence in your comments, each with an owner, a metric and a way to see that it worked. Action Plans of a Report are available in every package. See Create an Action Plan.
To run the same analysis on the next survey wave, create a new Report from the new file with the same choices. A Space keeps the feedback of a channel, such as a survey, and takes each new wave as a file, so the comments can be followed over time in one place. Spaces are available in the Enterprise package.
For survey data that is not an NPS survey, see How to Analyze Open-Ended Survey Responses. For more on what Kimola does with survey data, see the survey analysis page on kimola.com.