What are detractors actually complaining about?
Golden reshapes the survey, scores NPS, finds the weakest question, and sorts every comment by theme and sentiment, so 2,000 comments become two clear complaints.
2,000 comments and a deadline
Priya owns customer research. Twice a year she runs a satisfaction survey: ten agree-to-disagree questions, a 0 to 10 recommend score, and a comment box.
The export comes out one row per respondent with a column per question, which is easy to collect and miserable to chart. And the product team wants to know what detractors are saying, which used to mean a weekend with the comments.
Reshape, score, and read at scale
On upload, Golden notices the ten question columns are answers to one survey and offers to reshape them into rows. One click, with a before and after preview.
Then Analyze Text on the comment column adds sentiment and discovered themes to every response in one pass. Priya goes from raw export to a full readout in an afternoon.
How Golden answers it
Reshape the questions
Accept Golden's suggestion (or use Reshape Columns to Rows) to turn ten question columns into a Question and a Response. 2,500 respondents become 25,000 tidy rows.
Read the comments
Open the comment column's menu, choose Analyze Text, and pick Sentiment and Themes. Blank comments are skipped, and the new columns save with the dataset.
Find the weakest question
A diverging bar centers the 1 to 5 scale, with disagreement to the left. Sort by agreement and Support responsiveness lands at the bottom.
Track NPS by wave
A 100% stacked bar splits each wave into promoters, passives, and detractors, with a big-number tile for each wave's NPS.
Hear the detractors
A word cloud of themes filtered to detractors shows the two big complaints. A table of verbatims, one row per respondent, gives the product team quotes to read.
What the product team hears
- 01Support responsiveness is the weakest question: a 2.84 mean, with 37.8% of respondents disagreeing
- 02NPS climbs from +9 in Q1 to +22 in Q2, as promoters rise from 35.7% to 41.9%
- 03Of 488 detractor comments, 241 mention pricing and 180 mention slow or timed-out reports
- 04Sentiment runs negative for detractors and positive for promoters, a quick sanity check on the read
Keep everyone in their tools
Run Explain Change on NPS between waves to see where the lift came from (more promoters, mostly). Then generate a data story that quotes the weakest question and the top detractor themes.
Export the reshaped table to Google Sheets so the survey team keeps their pivot tables while the dashboard stays live. You can even skip the dialog and type "unpivot Q1 through Q10" in chat.
Hear every customer in one afternoon
Sign up for early access and turn your next survey into a readout.