Overall NPS is -11 with 68% renewal intent, but the segment gap is the real finding: Enterprise customers average 3.26 while Startup customers average 3.75. Longer-tenured customers score consistently higher — onboarding, not the product, looks like the pressure point.
Live render — every number above was computed from the sample dataset, and the same engine rebuilds this structure from your file.
How to use it: open the template with sample data, see how the blocks are built, then change the dataset to your own export — the engine regenerates the same structure from your numbers.
A survey report turns raw responses into a document that answers the question the survey was run to answer. The version people act on leads with the single most important finding, then supports it: a methodology box, the response overview, the closed-ended results, the themes from open-ended answers, and clear recommendations. The example above shows the shape on customer-survey data — responses, average satisfaction, NPS, and renewal intent up top, with satisfaction by segment and the finding written out.
Below: the anatomy of a survey report, how to present Likert and NPS results without distorting them, and phrasing you can adapt for the findings section.
The anatomy of a survey report
A survey report is not a chart per question in survey order — that's a data export, not a report. Structure it around what you learned, leading with the headline finding and organizing the middle by theme rather than by question number.
Six parts do the work: an executive summary with the one-sentence answer, a methodology box so readers can judge how much to trust it, the response statistics, the closed-ended results grouped by theme, the open-ended themes with representative quotes, and recommendations naming what the team will do next. The methodology box is small but load-bearing — it's what separates a credible report from an anecdote.
The sections of a survey report
Section
What it contains
Executive summary
The single most important finding, in one or two sentences
Methodology box
Sample size, response rate, dates, how respondents were reached, the instrument
Response statistics
Total responses, completion rate, and any segment breakdown of who answered
Closed-ended results
The rating and multiple-choice questions, grouped by theme with charts
Open-ended themes
Recurring themes from free-text answers, each with a representative quote
Limitations
What the survey could not measure — sample bias, self-selection, timing
Recommendations
What the findings mean the team should do
Always show the base (n) next to a percentage. "68% would renew" from 25 responses is a very different claim than from 400. A percentage without its base is the most common way survey reports mislead.
How to present Likert and NPS results
Rating-scale questions get mishandled more than any other type. Two rules keep them honest.
For Likert scales (strongly disagree to strongly agree, or 1-5 satisfaction), report the distribution, not just the average. A 3.4 mean can hide a badly polarized room — half love it, half hate it — that a single number erases. The clean summary is top-two-box: the share who picked the two most positive options ("agree" plus "strongly agree"). Show that alongside the mean and, where it matters, the full distribution as a stacked bar.
For NPS, the score is the percentage of promoters (9-10) minus the percentage of detractors (0-6); passives (7-8) count toward the base but not the score, which is why NPS ranges from -100 to +100. Report the score, but also the three group sizes — a +20 built on many promoters and many detractors needs a different response than a +20 where almost everyone is a passive.
Segment the results, don't just total them
The average across all respondents is usually the least interesting number in the report. The finding is almost always in the gap between segments — new versus tenured customers, enterprise versus SMB, one region versus another. In the example above, the overall NPS matters less than the spread between segments and the fact that longer-tenured customers score higher, which points at onboarding rather than the product. Always cut the key metric by your most important segment and lead with the gap.
Writing the findings: phrasing that works
The findings section is where reports go to die in a list of percentages. Three habits fix it, with illustrative phrasing (invented numbers, not real data):
Combine related numbers into a claim. Instead of "45% chose A, 32% chose B, 23% chose C," write "a clear majority (77%) preferred an option other than the current default." Lead each finding with the takeaway, then the number. "Onboarding is the weak point: satisfaction among customers in their first month averaged 3.1 out of 5, a full point below everyone else." Use active voice and plain language. "We surveyed 210 customers" beats "210 customers were surveyed." Each finding should be paired with one chart or one quote — never a wall of both.
Free-text answers are where the 'why' behind the ratings lives, and they're the most-skipped part of a survey report because they don't summarize themselves. The method is coding: read a sample, name the recurring themes (say, 'pricing,' 'support speed,' 'missing feature X'), tag each response, then report each theme with its rough frequency and one representative verbatim quote.
Don't cherry-pick the most dramatic quote — pick one that represents the theme. And report the themes that recur, not the single memorable rant, unless that rant points at something the numbers confirm. A theme like "14 of 60 comments mentioned slow support, echoing the low satisfaction score for that segment" ties the qualitative and quantitative halves together, which is what makes a survey report persuasive.
Mistakes that weaken a survey report
The reports that get politely ignored tend to repeat the same errors.
Leading with methodology. Sample sizes and dates belong in a box, not in the opening paragraph. Lead with the finding.
Reporting every question. The survey asked 30 questions; the report covers the 6 that matter. The rest go in an appendix.
Percentages without the base. Always show n. A percentage of a tiny sample is a headline waiting to mislead.
Averaging away the story. A single mean hides polarized or segment-specific results. Show distributions and cut by segment.
No recommendations. A finding with no 'so what' is trivia. End each theme, and the report, with what to do about it.
From responses to a finished report
Export your responses as CSV or Excel — Google Forms, Typeform, SurveyMonkey, and Qualtrics all do this — and upload the file. The report computes the response count, the satisfaction and NPS figures, the segment comparisons, and drafts the findings from the actual numbers, with the code behind each statistic attached so 'how did you get this?' has a real answer. Nothing on the page is illustrative; every figure is calculated from your responses.
It opens live in the workspace, regenerates when you upload a new wave, and ships as a share link plus a print-perfect PDF. Your file is analyzed in the browser and never stored on the server. For the analysis side end to end, see survey analysis and the guide on how to analyze survey data.
Frequently Asked Questions
Everything you need to know about using AnalyzeData.
The response overview, the key distributions, how segments differ, and what it means. This template renders all four from real sample responses — every statistic computed, not illustrative.
Google Forms, Typeform, SurveyMonkey, Qualtrics — anything that exports CSV or Excel uploads directly.
It depends on how precisely you need to read the results and how many segments you'll cut. For a rough directional read on a whole population, a few hundred responses is common; to report reliably on a segment, you need enough responses within that segment specifically, not just overall. The honest move is to report the base for every number and add a limitations note when a segment is thin rather than presenting a shaky percentage as solid.
Show the score — percent promoters (9-10) minus percent detractors (0-6) — and the size of all three groups, including passives (7-8). The score alone hides its own composition, so a reader can't tell a healthy +30 from a fragile one. Pair it with the open-ended 'why did you give that score?' themes, which is where the actionable detail lives.
Not in the main body — put the full question-by-question breakdown and any raw exports in an appendix, and keep the report itself focused on findings. Readers who want to verify a number can go to the appendix; the rest are served by the summary and the themes. Including everything up front is the fastest way to get a report skimmed and set aside.
Code them into themes rather than quoting at random. Read a sample, name the recurring topics, tag each response to a theme, then report each theme with its frequency and one representative quote. Tie the themes back to the quantitative results where they overlap — a theme that explains a low rating is far more useful than a list of disconnected comments.
Build your survey report now
Open the template with sample data, then swap in your own file.