A survey report turns a pile of answers into a document someone can act on. It leads with the answer to the question you asked, then covers the method, who answered and who didn’t, themes with how many people raised each, quotes in people’s own words, where groups disagreed, and what to do next. A survey results report is the same document under another name. Example: “Of the 64 people who answered, 27 said priorities change mid-week; first change: set them once, on Monday.” Below: the structure, an example, how to present survey results by question type, the mistakes, and what to ask yourself before you write.
One exercise carried the course. In the report: a theme, the number of people who named that exercise, and this quote next to it.
In this order. A reader with two minutes gets the answer from the first section; a reader with twenty gets the evidence behind it.
The Agent Results Brief for an employee survey, laid out the way this page describes: the answer at the top, themes with the number of people behind each, where two groups disagree, and quotes from the biggest theme; in this example the actions have been written in as “What to change first.” Sixty-four of eighty-eight invited people answered. The questions behind it are on the employee feedback page; customer feedback has a second example.
People like the work and the colleagues around them, and that is the strongest thing here. What wears them down is priorities that change during the week and small changes waiting on approval. Change first: set priorities once a week and say what gets dropped, then shorten sign-off for small changes.
An example of the Agent Results Brief. The people, counts and quotes are illustrative.
Each question type has one honest way to show it and one or two tempting ways that mislead. The counts stay in; the decoration goes.
| Question type | Show | Leave out |
|---|---|---|
| Yes/no and multiple choice | The count for each option as a bar, and the share only when enough people answered for a share to mean something | Pie charts with six slices; percentages on a dozen answers |
| 1–5 scales | The spread: how many picked each point, next to the average | The average on its own (a 3 can be everyone at 3, or half at 1 and half at 5) |
| NPS 0–10 | Promoters, passives and detractors as counts, then the score, then the reasons people gave | The score with no reasons; see NPS follow-up questions |
| Open-ended | Themes with counts, one or two quotes each, and the disagreements | Word clouds; a list of every answer; a single quote standing in for a theme |
| Ranking | What most people put first, and what almost nobody did | A weighted average rank, which no reader can picture |
| Any question, by group | The same count side by side for the groups that differ | One combined number when the groups pull in opposite directions |
| A repeat survey | This run’s count next to last run’s for the same question | A trend line with no counts behind it |
Whatever the type, say how many people the number rests on before you say the number.
Written answers are often the most useful part of a survey, and the hardest to report. Three moves turn sixty paragraphs into a page a reader can act on. The questions that produce good written answers are in open-ended survey questions.
Read every answer once, then tag each with the one thing it’s about, in the words people used. Merge tags that mean the same thing. Stop at three to five themes; a sixth is usually two answers.
Say how many people raised each theme, out of how many answered. A count is what separates “some people mentioned” from something a reader can weigh against the other themes.
Pick the quote that says the theme plainly, not the angriest or the most articulate. Keep it whole. Put the quote next to its theme and its count, so the reader sees the evidence and the scale together.
When one group says a thing helps and another says it hurts, the average says “neutral” and the reader learns nothing. Show the split.
Many first answers are one line and polite. The quote that carries a theme usually comes from the follow-up. Two examples, neither from a real survey.
One exercise carried the course. In the report: a theme, the number of people who named that exercise, and this quote next to it.
“Mostly clear” hid one specific handoff nobody announced. That goes in the report as a quote beside its theme, and as an action with an owner.
Each one is easy to make on a Friday afternoon and easy to spot in someone else’s report.
“42% of respondents” sounds like evidence until the reader learns it means five of twelve. Below a few dozen answers, report counts; and always say the count first.
Three quotes that agree with you read as a case, not a finding. Choose quotes by how plainly they say the theme, and include one that disagrees.
A report that stops at “here’s what people said” hands the hard part to the reader, who won’t do it. End with one to three actions, an owner and a date.
It shows which words were common, not what anyone meant. “Time” is big in the cloud whether people want more of it or less. Themes with counts replace it.
Method, demographics, then a chart per question, then the findings. A busy reader rarely gets past page two. Put the answer in the first paragraph and the method after it.
A 3.1 out of 5 can mean everyone is lukewarm or that two groups are at war. Show the spread, and where groups disagree, show both sides with counts.
Twenty-two questions in four groups. Answer them in a scratch file first; the report almost writes itself from the answers. Under each is the follow-up to ask yourself when your first answer is vague, which it often will be.
What question was this survey meant to answer?
Write it in one sentence. If it takes two, which one does the report answer?
What is the one-sentence answer?
Which theme, with its count, is that sentence resting on?
What decision will the reader make with this?
What does the reader need to see on the first page to make it?
What did you expect to find before the answers came in?
Which finding is furthest from what you expected?
What would the reader do differently on Monday?
What in the findings points at that change?
How many people answered, out of how many invited?
Where in the report is that fraction shown before any percentage?
Who didn’t answer?
What do the people who didn’t answer have in common?
Which groups answered more than others?
Which finding would change if the quiet group had answered?
How were people asked?
What in the wording or the timing could have shaped the answers?
How was anonymity handled?
Which answers might people have softened because of who could read them?
What are the three to five themes?
How many people raised each one? Say the number, not “many.”
Which theme has the most people behind it?
Which quote from that theme says it in a way no summary could?
What is working, not only what’s broken?
Which theme would people miss if it changed?
Where did people disagree?
How many people are on each side?
Which quotes are you using?
Which quote, if any, cuts against your answer?
What surprised you?
Where in the report does that surprise appear?
What can’t this survey tell you?
Which of your claims leans on something the survey didn’t ask?
What should happen first, as a result of this?
Who owns that first action?
What will you tell the people who answered?
Which of their themes will they see acted on?
When will you ask again?
Which number would you expect to move by then?
What’s the one thing you’d cut from this report?
What would the reader lose without it?
How long is the report?
What could move to an appendix so the answer is on page one?
For a survey of a few dozen to a couple of hundred answers, written up by one person. Longer surveys take longer in the reading, not in the writing.
Every answer, start to finish, without tagging. You’re looking for the shape: what keeps coming up, what surprised you, who sounds different from everyone else.
Second pass: one tag per answer, in the words people used. Merge near-duplicates. Count each tag. Note the groups that pull in different directions on the same question.
Two or three sentences at the top: what you asked, the biggest theme with its count, what to do first. Then the method and who answered, from your notes. Then the themes, biggest first.
Pick the quotes. Lay out each disagreement as two counts. Write one to three actions with an owner and a date. Read the whole thing once as the busiest person who will get it.
With a classic survey, the afternoon above is yours. With an agent survey, Spoke interviews each person and the Agent Results Brief does the reading, tagging and counting; you check it against the transcripts and write the actions.
A document that turns survey answers into something a reader can act on: the answer to the question you asked, how you asked it, who answered, the themes with counts, quotes, where groups disagreed, and what to do next.
Read every answer once, tag and count the themes, then write the answer first: two or three sentences at the top. Follow with the method, who answered, the themes with counts and a quote each, the disagreements, and one to three actions with owners. Most reports of a few dozen answers fit in an afternoon.
Seven sections: the answer, the method, who answered (and who didn’t), themes with the number of people behind each, quotes, disagreements between groups, and what to do next. Put the questions themselves in an appendix.
By question type. Multiple choice: counts, and shares only when enough people answered. Scales: the spread, not just the average. Open-ended: themes with counts and a quote each. Where groups disagree: both sides with their counts, side by side.
Theme, count, quote. Tag each answer with the one thing it’s about, merge tags that mean the same thing, stop at three to five themes, count how many people raised each, and put one plain quote next to each theme. Skip the word cloud.
As long as the evidence needs and no longer: for most surveys, two to four pages, with the answer in the first paragraph and the question list, tables and extra quotes in an appendix. A one-page version for busy readers is often the one that gets read.
Results are the raw output: counts per option, the list of written answers, an average per question. A report interprets them: what the results mean, how far to trust them, where people disagree and what to do. Results are for the person who ran the survey; the report is for everyone else.
The same order, one section per slide: the answer, who answered, one theme per slide with its count and one quote, the disagreement as two counts, the actions. Leave the method and the full question list in a handout or an appendix slide.
Spoke asks the follow-ups for you, with every person. Free to start, no card. Your first ten agent surveys are on us.
