The discussion section sits between the findings and the conclusions in a business report, and it's the section writers get wrong most often. They either repeat what the findings already said, or they jump straight to what should be done, skipping the analysis that earns the reader's trust. A discussion section explains what the findings mean, not what the findings are.
What a discussion section is actually for
Findings present data. Conclusions summarise the verdict. The discussion section does the interpretive work in between. It answers the reader's real question: so what?
In a well-structured report, the discussion section takes each finding and explains its significance in context. It notes where findings align with expectations and where they don't. It acknowledges limitations. It draws connections between pieces of data that, read separately, wouldn't tell the full story. Without this section, a report is a data dump with a verdict attached. With it, the reader can follow your reasoning step by step.
This section sits logically after the findings section, which presents evidence, and before the conclusions section, which declares the overall position. The three sections form a chain: evidence, meaning, verdict. Cut the middle link and the chain breaks.
How to structure a discussion section
Start with the finding that carries the most weight. Don't save the most important point for last, because busy readers often don't reach the end. Open the discussion with your most significant interpretive move, then work through the remaining findings in order of importance.
For each finding, follow this three-part structure:
- State what the finding shows. One sentence, no more. Don't re-read the data table aloud; identify the pattern or result.
- Explain what it means in context. This is the actual analysis. What does this result imply? How does it compare to a benchmark, a prior period, or a target? Why is the number high or low?
- Connect it to the report's central question. Every finding should tie back to the reason the report was commissioned. If you can't connect it, the finding doesn't belong in the discussion.
Repeat this pattern for each finding. The section shouldn't be a single long argument. It should read as a structured walk through the evidence, with each stop building the case.
The most common mistakes in discussion sections
Writers most frequently copy data verbatim from the findings section. The sentence "Staff satisfaction fell from 72% to 61%" is a finding. "The 11-point drop in staff satisfaction over 12 months suggests the restructure has not restored confidence at the team level" is a discussion. The difference is interpretation. If your discussion sentences could have appeared unchanged in the findings section, rewrite them.
A second mistake is hedging every sentence. Phrases like "this may possibly suggest" or "it could perhaps indicate" empty the section of meaning. Commit to the interpretation the evidence supports. If you're genuinely uncertain, note the limitation clearly: "Without data from the previous financial year, it's difficult to determine whether this pattern predates the policy change." That's a real qualification. "This could mean something" is not.
Third: ignoring contradictory findings. Real data rarely tells a perfectly clean story. If two findings pull in different directions, the discussion section is exactly where you address that tension. Acknowledge it, offer the most plausible explanation, and note what further data would resolve the question. Readers trust writers who engage with complexity more than writers who cherry-pick the easy results.
Discussing limitations without undermining the report
Every report has limitations. Sample size, data collection timeframe, self-reporting bias, scope constraints. Raising these in the discussion section isn't a weakness. It's good analysis. The key is to state the limitation, explain its effect on the findings, and then move on. Don't dwell on it or apologise for it.
One sentence on a limitation is usually enough. "The survey covered 43 respondents from the Sydney office only; results may not reflect conditions across the Brisbane and Melbourne teams." That's clear, honest, and bounded. The reader knows what the data can and can't tell them. This kind of precision is exactly what separates a credible report from an overclaiming one.
Connecting the discussion to conclusions and recommendations
The discussion section does the heavy lifting so the conclusions section doesn't have to. When the conclusions section reads clearly and confidently, it's usually because the discussion section did its job. If you find yourself over-explaining in the conclusions, the discussion probably left interpretive work undone.
A good test: read your discussion section and then read your conclusions section. The conclusions should feel inevitable given what the discussion established. If the conclusions introduce ideas that weren't set up by the discussion, the structure has a gap. Fix it in the discussion, not in the conclusions. On a related note, the recommendations section works best when it draws directly from conclusions that the discussion has already made credible.
Tone and length
Keep the tone analytical, not conversational. The discussion section isn't the place for enthusiasm ("the results are encouraging!") or alarm ("this is a serious problem"). State the interpretation plainly and let the evidence carry the weight.
Length depends on the complexity of the findings. A report with 4 distinct findings might have a discussion section of 300 to 500 words. A complex report with 10 findings across multiple variables might run 1,000 words or more. What matters isn't word count. It's that every paragraph moves the interpretation forward. If a paragraph doesn't add a new interpretive point, cut it.
Don't begin the section with "In this discussion section, we will analyse..." Start with the first substantive point. Readers know what the section heading says.
A quick example
Weak: "Customer response times increased from an average of 4.2 hours in Q1 to 6.8 hours in Q3."
Stronger: "The 62% increase in average customer response times between Q1 and Q3 aligns with the headcount reduction in the client services team in late Q2. The data suggests that the team is managing a higher ticket volume per agent without additional support, which is creating the delay. This directly affects the 4-hour response target in the current service-level agreement."
The second version does four things the first does not: it calculates the magnitude of change, identifies a plausible cause, explains the mechanism, and connects the finding to something that matters to the organisation. That's what discussion sections are for.