"We ran the survey, wrote the report, and nobody read it." A textbook organizational outcome.
Report templates are everywhere online; articles about why reports get read or ignored are rarer. This article covers structure / charts / page count, plus the substance: designing reports for the reader's decision.
7 principles for a read report
Principle 1: Always include a 1-page summary
Long reports don't get read in full. The opening 1-pager should contain:
- Survey purpose / period / response count
- 3 key findings
- 3 proposed next actions
Decision-making must be possible from just that page.
Principle 2: Write "number → interpretation → recommendation" as a set
✗ Number alone: "NPS was 32"
○ Set: "NPS 32 (up 4 from last period, 12 below industry average).
Improving but not at industry level.
Consider plays to lift passives."
Numbers alone don't drive action. Interpretation + recommendation lets the reader move.
Principle 3: Always include comparisons
Numbers without comparison don't mean anything. At minimum compare with one of:
- Prior survey
- Year-on-year
- Industry average
- Target
- Cross-segment
Principle 4: Quote representative open text
Number-only reports feel cold. Quoting 3–5 representative open-text responses adds emotional temperature and engages the decision-maker.
Principle 5: Pick charts by purpose, not type
When in doubt, pick "easiest for the reader to understand." Cap at 3–4 chart types:
| Purpose | Recommended chart |
|---|---|
| Distribution | Pie / bar |
| Trend | Line |
| Cross-category | Horizontal bar |
| Distribution shape | Histogram |
| Multi-item summary | Scorecard (number table) |
Fancy charts (radar, bubble) raise cognitive cost for readers — avoid.
Principle 6: Mind your page count
| Purpose | Recommended pages |
|---|---|
| Exec staff summary | 1 |
| Department-shared report | 5–10 |
| Detailed analysis | 20–30 |
| Detailed data dump | Separate (Excel) |
Different versions per purpose is the ideal. Trying to cover everything in one doc satisfies no one.
Principle 7: Close with next actions
End with "what's next":
[Next actions]
1. Review support SLA (owner: Tanaka [CS], by end of May)
2. Individual interviews with churn-risk segment (owner: Sato [Sales], by 5/15)
3. Next re-survey: 2026-07-30
Names and dates turn the report from "meeting material" into "an execution plan."
Standard report structure
For a mid-scale survey (200–500 responses), a 10-page structure:
1. Executive summary (1 page)
- Survey overview
- 3 key findings
- Proposed actions
2. Methodology (1 page)
- Purpose / context
- Period
- Audience / response count / response rate
- Question breakdown
- Caveats
3. Overall results summary (1–2 pages)
- Key numbers (NPS, CSAT, satisfaction)
- Comparison to past
- Comparison to industry
4. Segment trends (2–3 pages)
- Numbers by industry / scale / tenure
- Anomalous segments
- Cross-segment comparison
5. Open-text analysis (2–3 pages)
- Category aggregates
- Representative-voice quotes per category (3–5 each)
- Comparison to past
6. Key findings and hypotheses (1–2 pages)
- Facts visible in numbers
- Interpretation / hypotheses
- Open questions for additional probing
7. Proposed actions (1 page)
- Short-term (within 1 month)
- Mid-term (3–6 months)
- Owners / deadlines / success definitions
8. Detailed data (separate Excel)
- All question aggregates
- Per-segment cross-tabs
- Full open text
The hard part — structural causes of "won't get read"
Cause 1: Ignoring the reader's time budget
Execs / department heads want to digest a report in 3–5 minutes. Authors want to "show their work" with detail.
Counters:
- 1-page summary at the top, always
- Detail in appendix
- "Designed for the busiest reader" mindset
Cause 2: Data dump with no conclusion
"Here are the numbers." "Here are the voices." With no conclusion, the reader has to derive it. High cognitive cost → unread.
Counters:
- Always write "so what" after data
- Don't fear sharing your own interpretation
- Drop the "the reader will interpret" assumption
Cause 3: Too many charts
"More charts = easier" is wrong. Readers pay per-chart reading cost — the whole picture becomes invisible.
Counters:
- 3–5 most important charts only
- Rest as number tables
- Avoid decorative charts
Cause 4: Open text dumped in full
"We included every response" looks generous but doesn't get read.
Counters:
- Classify first, then quote representatives
- Full dump in appendix
- Main report includes only voices meaningful for the reader
Cause 5: No "what's next"
Reports ending at "here's the data" stop at meeting material.
Counters:
- Mandatory proposal section
- Names and dates
- Success definitions
Self-review checklist
After finishing, run through:
- 1-page summary captures the overview?
- Key numbers have comparison reference?
- Numbers paired with interpretation and recommendation?
- Representative open text quoted?
- Charts ≤ 3–5 types?
- Total volume matches purpose?
- Next actions have owners and deadlines?
- Readable in 5 minutes?
- Structured so the reader can enter the discussion?
Pass these and report quality stabilizes.
Version per stakeholder
Same survey, different formats per audience — multiplies impact.
Exec staff / board
- 1-page summary
- 3 directly board-actionable points
- Weight on "what to decide" over numbers
Department head / manager
- 5–10 pages
- Their area's segment detail
- Action list
Field practitioners
- Detailed data (Excel)
- Full open-text access
- Material for improvement debate
All-staff
- Excerpt (for transparency)
- Headline numbers and direction
- As a feedback letter
Respondent feedback
- 1-page light note
- Focus on "how we'll act on your input"
- Build participation expectation for next round
Notes on ChatGPT / Copilot-assisted reports
AI-drafted reports are increasing. Cautions:
- AI tends to be sloppy on interpretation
- Numerical hallucinations are frequent
- Industry context / company specifics don't transfer
- AI for draft + human for interpretation + recommendation is the best workflow
"AI-written, ship it" reports get detected by readers — credibility drops.
Where Repoan fits
Repoan ships features designed for "writing a read report":
- Auto 1-page summary — AI generates summary right after collection
- Comparison vs. past surveys — auto-displayed
- AI open-text categorization — themes with representative-voice extraction
- Segment dashboard — one-click switching
- PDF and Excel export — pre-formatted reports downloadable
- AI interpretation text — initial comments on numbers, edited by humans
Summary
Writing reports that get read:
- 1-page summary at the top
- Number → interpretation → recommendation as a set
- Always include comparisons
- Quote representative open text
- 3–5 chart types max
- Different versions per purpose
- Close with next actions
The goal isn't "produce a report" — it's "trigger a decision." Designing around the reader's time budget and cognitive load is what makes a report not get archived.