The most careful analysis is wasted if nobody understands it. Explaining data clearly is a skill of its own, and it is one that non-technical professionals are often naturally good at, because they know what their audience cares about.
Lead with the answer
Start with the conclusion, then support it. Instead of walking through every step you took, say: "Online sales fell 8% last month, mainly because returning customers bought less." Then show the evidence.
Use the "what, so what, now what" structure
- What: the finding. "Returning-customer orders dropped from 520 to 410."
- So what: why it matters. "Returning customers usually make up 60% of revenue."
- Now what: the recommendation. "Restart the monthly email offer we paused in May and track results for four weeks."
Translate the jargon
| Instead of | Try |
|---|---|
| "Statistically significant uplift" | "A consistent improvement, not just a lucky week" |
| "CTR decreased 0.4 pp" | "Fewer people clicked the ad after seeing it" |
| "Negative correlation" | "When prices went up, orders went down" |
| "Outliers skewed the mean" | "A few very large orders made the average look higher than normal" |
Choose one chart per message
Each chart should make one point. Give it a title that states that point ("Returning-customer orders fell in June") rather than a generic label ("Orders by month"). Highlight the important bar or line in a strong color and keep everything else neutral.
Avoid misleading visuals
- Start bar chart axes at zero.
- Do not cherry-pick date ranges to exaggerate a trend.
- Label units clearly (dollars, percentages, counts).
- Mention data limitations openly.
Use relatable comparisons
"$12,000 in lost revenue" may be abstract. "Roughly the cost of our whole spring campaign" lands immediately with a marketing team.
Prepare for questions
Anticipate two or three follow-ups ("Is this seasonal?", "Which products?") and have a backup slide or table ready.
Data storytelling is the focus of Week 4, which ends with the final mini-project.
