Trend over time
Line chartShow direction, seasonality and turning points.
Analysis · Module 06
Visualization & Storytelling.
A dashboard is useful only if someone can look at it, understand what changed, and know what deserves attention next.
Core idea
The job of a chart is not to show all the data. It is to make the important comparison hard to miss.
01
Context
The executive dashboard has 27 charts.
“The dashboard has everything, but we still spend the first 20 minutes asking what actually changed.”
That is a visualization failure. A dashboard can be technically complete and still make the decision harder.
Your goal is to create information hierarchy: what changed, why it matters, where it happened, and what deserves investigation.
02
Chart choice
Match the chart to the analytical task.
Compare categories
Bar chartMake ranking and magnitude easy to scan.
Part-to-whole
100% stacked barUse only when composition is the question.
Distribution
Histogram / box plotShow spread, skew and outliers.
Relationship
Scatter plotExplore association between two numeric variables.
Single KPI
Number + contextPair the value with target, change or comparison.
Rule
Choose the chart after you know the comparison. “I want a pie chart” is not an analytical requirement.
03
Dashboard hierarchy
Design from decision to detail.
Weekly business review
-11.9% Orders fell because cancellations rose in 2 cities.
Completed orders63.4k-11.9% WoW
Cancellation rate15.1%+6.9 pp
Sessions516k+3.2% WoW
New users86k+4.9% WoW
Trend
What changed
Dhaka and Chattogram explain most of the cancellation-rate increase. Top-of-funnel traffic remained healthy.
The dashboard starts with the answer hierarchy: headline → primary KPIs → trend → driver. Filters and diagnostic detail can come later.
04
Storytelling
Write the insight into the visual.
Cancellation Rate by City
Describes the chart, but makes the reader do the analytical work.
Dhaka and Chattogram drove 72% of the cancellation increase.
States the relevant comparison and focuses attention immediately.
Writing rule
Use the title for the conclusion, the subtitle for context, and annotations for evidence.
05
Challenge
Cut the dashboard in half.
Exercise · 15 minutes
Keep only what earns its space.
- If the decision is whether to escalate an operations issue, which 4–6 visuals are essential?
- Which charts are diagnostic and can move to a drill-down page?
- Which KPI needs comparison context rather than a raw number?
- What should the page headline say if cancellations are the main driver?
- Where should data-quality warnings appear?
- What would you remove even if a stakeholder originally requested it?
Reveal one possible dashboard structure
01Headline
One sentence explaining the main business change.
02KPI strip
Completed orders, cancellation rate, demand, conversion.
03Trend
Show when the shift began and whether it persists.
04Driver view
City or segment contribution to the change.
05Diagnostic drill-down
Reasons, platform, customer type, operational detail.
06Data note
Freshness, exclusions and confidence caveats.
06
AI assist
Use AI as a dashboard critic.
Visualization review prompt
You are reviewing a weekly operations dashboard.
Decision:
Should leadership escalate the recent completed-order decline to Operations?
Known analysis:
- completed orders -11.9% week over week
- sessions +3.2%
- new users +4.9%
- cancellation rate increased from 8.2% to 15.1%
- Dhaka and Chattogram explain most of the increase
- data-quality checks have passed
Review the dashboard design:
1. Identify the 4–6 visuals that should be on the first page.
2. Explain what decision each visual supports.
3. Suggest an insight-led title for each.
4. Identify any misleading or redundant chart choices.
5. Suggest what should move to drill-down pages.
6. Do not invent additional findings. 07
Critique
Questions every dashboard should survive.
01
What decision should this visual help someone make?
02
What is the one main message?
03
Is the chart type appropriate for the question?
04
Are the axes, scales and time windows honest?
05
Is comparison context visible?
06
Can a viewer distinguish signal from noise quickly?
07
Does the title state the insight rather than the chart type?
08
Is uncertainty or data-quality risk visible when relevant?
08
Worked communication
Lead with the decision-relevant story.
Executive summary
“Completed orders fell 11.9% week over week even though sessions and new users increased. The decline is primarily associated with higher cancellations, concentrated in Dhaka and Chattogram. The next action is to investigate cancellation reasons and operational capacity in those two cities rather than reduce acquisition spend.”
Say what changed and what appears to drive it.
Use a small number of visuals that directly support the claim.
Show comparison periods, definitions and relevant caveats.
End with the next decision or investigation, not a generic summary.