How to write a market research report, step by step

Step 1: Who will read the report, and what will they decide?

A report is written for a decision, not for a dataset. Before opening the survey tables, write one sentence in this shape: "By [date], [who] will decide whether to [action], and needs to know [what]." If you cannot fill in the gaps, the analysis will drift towards whatever is easiest to chart.

Then name the reader. A managing director wants the answer and the main risk in a few minutes; a product manager wants detail by segment; a finance team wants the assumptions behind any forecast. One document can serve all three only if its layers are clear: the summary for the first, the findings chapters for the second, the appendix for the third.

Throughout this guide we use a hypothetical case: a British firm selling social-media scheduling software to companies, deciding whether to open a Bulgarian or a Dutch sales desk first.

Step 2: Turn the decision into research questions

Break the decision into four to six questions that evidence can answer. For the software firm they might be:

• How many companies in each country already use social media for business?
• How many pay for online advertising, and is that share growing?
• What slows down marketing teams that publish on social media?
• Would they pay for a tool, and through which buying process?

Each question becomes a chapter, and each chapter starts from a research hypothesis that the data will confirm or reject, for example "adoption in Bulgaria is too low to support a local desk". Writing hypotheses down early stops you collecting interesting facts that answer nothing. If a research brief already exists, reuse its wording, so the client recognises the questions they commissioned.

Step 3: Audit the evidence before you write a word

List what you hold against each question: source, period, definition, base size and how far it can be trusted. This evidence log takes an hour and saves days of rewriting, because gaps surface before the draft rather than in the client meeting.

Official series like these come from desk research; attitudes and intentions need fieldwork such as surveys or interviews. For every survey figure, record the base and the margin of error in the log, so the strength column rests on arithmetic rather than instinct.

Step 4: Build the storyline and the skeleton

Write the storyline before the chapters. Put the answer in one sentence, add the three or four findings that support it and list the evidence under each. Consultants call this a pyramid; in practice it is a one-page outline the client could approve in a short call.

Next, write the headline of every section and read the headlines in sequence, without the body text. If they tell the story on their own, the structure works. If they read "Background", "Social media", "Advertising", you have labels, not findings.

The skeleton itself is conventional: summary, background and objectives, method, findings chapters, conclusions and recommendations, appendices. Our market research report format guide sets out the full table of contents and length guidance. The rule that matters here is order: findings appear in order of importance to the decision, never in questionnaire order.

Step 5: Draft each section as claim, evidence, implication

Every findings section follows one pattern. The headline states the claim, the first paragraph proves it with the strongest evidence, a chart or table shows it, and the closing sentence says what it means for the decision. Someone who skims only headlines and closing sentences should still follow the argument.

Writing rules that keep the draft honest:

• Give every number its base, period and source. "41.5% of enterprises with 10 or more employees (Eurostat, 2025)" can be checked; "about 40% of companies" cannot.
• Round consistently. One decimal for official percentages, whole numbers for survey percentages on a base of a few hundred.
• Write differences in percentage points. A rise from 20.9% to 25.6% is 4.7 points, not 4.7%.
• Separate fact from interpretation. Signal the move from data to judgement with phrases such as "this suggests" or "we conclude".
• Keep qualitative findings qualitative. Twelve interviews support "most managers mentioned", not a percentage.
• Use the client's vocabulary. If they say "accounts", do not switch to "customers" in chapter four.

Step 6: Choose visuals that prove the headline

A chart earns its place when it proves the headline above it faster than a sentence could. Pick the form from the message, not from the software's default setting.

The chart title above is the finding itself, not a description of the axis. Under every visual, give the base, the period and the source. Keep each country or brand in the same colour on every page, and start bar axes at zero so that differences are not exaggerated.

Step 7: Write the summary and recommendations last

Only when the findings chapters are stable do you know what the answer is, so the summary comes last. It opens with the recommendation, gives the three or four findings that support it and closes with risks and next steps. Our guide to the executive summary of a market research report covers its structure in detail.

Recommendations belong in their own chapter, numbered and tied to the findings that justify them. Each says what to do, who owns it, by when, and which result would show it was the wrong call. For the software firm that could read: open a Dutch desk in the first half of the year, run a pricing study before entering Bulgaria, and revisit the Bulgarian decision when Eurostat next publishes its figures on business use of social media.

Step 8: Check every figure, then hand it to a cold reader

Quality control is part of writing, not an afterthought. Before release:

1. Trace every number in the summary to a table, and every table to its source.
2. Check that percentages add up to 100, or explain why not, for instance with multiple answers.
3. Test any difference you call meaningful for statistical significance.
4. Confirm that definitions and periods match across chapters.
5. Record the extraction date of each official series, because statistical offices revise their data.
6. Ask someone who has not worked on the project to read the summary and tell you, in one sentence, what the client should do.

If that colleague names the wrong action, the summary needs rewriting, however accurate the data behind it.

Which mistakes weaken a market research report?

• Reporting the questionnaire. Question-by-question charts show effort, not insight. Group results around the decision.
• Burying the answer. If the recommendation first appears on page 40, many readers never reach it.
• Mixing sources in one chart. A national office's figure and a harmonised Eurostat figure can measure different things; keep them apart.
• Over-reading small bases. A subgroup of 40 respondents carries a margin of error of about ±15 points at 95% confidence; flag it or merge it with another group.
• Hiding the limitations. State what the study could not measure. It builds trust and prevents misuse of the findings.
• Letting AI tools invent sources. Drafting help is fine, but every figure still needs a verified source. Our piece on market research with AI explains where the line sits.