Data
Reading a Chart Honestly: Five Questions Before You Believe It
Charts persuade quickly. A short checklist helps you slow down and see what is really being claimed.
Nicole Jackson
Lead Data Analyst · Sep 4, 2026 · 6 min read
A well-made chart can end an argument. A badly made one can start the wrong one. Whether you are producing visuals or consuming them in a meeting, a few habits will protect your judgement.
The five questions
- 01Where does the axis start, and is that choice hiding or exaggerating change?
- 02What is the denominator? Raw counts rarely compare fairly across groups.
- 03What period is shown, and what would a longer window reveal?
- 04Who is missing from this data?
- 05What decision would change if this number were twice as large?
Why the last question matters
Analysis exists to inform choices. If no decision would change, the chart is decoration. If a single outlier would reverse your recommendation, you need more data before you present.
“A chart is an argument wearing a lab coat.”
Make your own charts easier to trust
Label directly, annotate the interesting moment, and write the conclusion as the title. 'Support tickets fell 18% after the new help centre launched' says more than 'Tickets by month'.
Put it into practice
Courses that build on this idea
Data Analysis with SQL
Query real datasets with confidence, from simple filters to window functions and cohort analysis.
Data Visualization That Speaks
Choose the right chart, remove the clutter and present data stories decision-makers remember.
Keep reading
All articles
Learning Science · 7 min read · Sep 29, 2026
Why Spaced Practice Beats Cramming (and How to Start Tonight)
The brain keeps what it has to work to retrieve. Here is how to rebuild your study routine around that single idea.
Design · 6 min read · Sep 21, 2026
The Case for Boring Interfaces
The best products rarely feel clever. They feel obvious, and obvious takes a lot of work.
Careers · 8 min read · Sep 13, 2026
Your First Year as a Product Designer: A Realistic Map
What to expect in months one, six and twelve, and what to do when the work does not look like your portfolio.
