Storytelling with data: turning analysis into reports and presentations people act on

An analysis only matters if someone understands it and acts on it. Managers are busy; they want the answer, why it matters and what to do. Storytelling with data means presenting your findings clearly, honestly and in the right order.

Start with the answer (the pyramid)

Most reports go: background → method → lots of tables → conclusion on page 9. Flip it:

1. The answer / recommendation (one or two sentences)
2. The key evidence (two or three findings with charts)
3. Details and method (appendix, for those who want it)

Weak opening: "This report analyses sales data from January to June 2026 across four branches…" Strong opening: "Thika branch sales fell 16% in April (KSh 75,000), mainly because two best-selling fertilisers were out of stock at planting time. Restocking three weeks earlier would recover most of it."

From data to insight to action

Always push past "what happened" to why and so what.

Structure of a one-page report

1. Title that states the finding
2. Summary: 3 bullet points
3. One or two charts that prove the main point
4. Recommendation(s) with expected impact
5. Next steps and who owns them
6. Notes: data source, period, limitations

Presenting to managers

• Know your audience: a board wants the big picture; a branch manager wants their numbers.
• One idea per slide, with a title that states the point.
• Highlight the bar or line that matters; grey out the rest.
• Use plain language: "about one in five customers", not "20.3% (n = 1,412)".
• Prepare for questions: have the details ready but don't show them all.
• Be honest about limitations: missing data, small samples, correlation vs causation.

Writing numbers well

Turn analysis into a summary (run it)

branches = {"Thika": (455000, 380000), "Kenol": (295000, 310000), "Nyeri": (338000, 342000)}

lines = []
for name, (march, april) in branches.items():
    change = (april - march) / march
    lines.append((change, f"{name}: {'up' if change > 0 else 'down'} {abs(change):.0%} (KSh {march // 1000}k → {april // 1000}k)"))

lines.sort()
worst = lines[0][1]
print("Headline: " + worst.split(":")[0] + " had the biggest fall in April.")
for _, text in lines:
    print(" -", text)

Summary

• Start with the answer, then the evidence, then details.
• Move from data to information to insight to recommendation.
• One idea per slide or chart; plain language; rounded numbers.
• Be honest about limitations; your credibility matters most.