Is my dashboard ready to share? Review it with five questions

Lucía asked her AI to fix the numbers, then to build the whole dashboard. Here's what she got.

Illustrative scenario: written for this course, not a real response from any AI.

Clínica Alba · Week of September 21, 2026

No-show rate 11.5% (+0.3% vs. last week; target 10%) · Occupancy 71.6% · Completed 331 · Wait 19.7 min

Bars of no-show rate by site, highest to lowest: Cholula 17.1%, Centro 10.1%, Angelópolis 9.8%, with a dotted target line.

Bars of occupancy by site.

Line of no-show rate by week, starting June 29, with a dotted target line.

Stacked bars of each site's no-shows by week.

It's much better than the first attempt. But better isn't ready. Before you read on, review it with the list below: two numbers and one chart shouldn't look like this.

Does it answer the purpose sentence?
From the first screen, does Lucía know which site gets the calls and whether there's room on the schedule? Anything that doesn't help with that comes out.
Does each number follow its definition sheet?
Formula, grain and period. Recalculate from the totals the AI gave you, as on the previous page.
Does each number come with a comparison?
Against last week and against the target. The difference between two rates is stated in percentage points; a % sign means a relative change.
One chart per question?
Each chart answers one question from the brief, and none repeats another. To compare sites of different sizes, use rates, not counts.
Does it read in the order of the decision?
The answer at the top, the where in the middle, the since-when at the bottom, with the period stated at the top.

Check yourself: What's wrong with "Occupancy 71.6%"?

Check yourself: The clinic's no-show rate went from 11.2% to 11.5%. How should the dashboard card say it?

The third problem is chart 4. It counts no-shows by site, and Centro, the biggest site, will always look worse even when its rate is on target. Besides, chart 1 already answers "where?" with rates. Chart 4 comes out.

Before you share

Once the dashboard passes the list, add three things that make people trust it:

A source and date line: where the data comes from and when it was last updated.
The definition sheets one click away: a tab or a link with the definition of each number.
Who to ask: a name, so a question doesn't turn into another dashboard.

And before you send the link, check who can see it. Also find out whether the dashboard refreshes on its own or whether next Monday you'll have to ask for it again with the new file. If it's the latter, keep the brief and the five questions next to the dashboard: they're what you'll repeat, so next Monday you don't start from scratch. If you want to review your own dashboard with this list, describe it to your Maestro.

The three opening questions

On the first page you answered three questions before you started. Here they are again, exactly as they were. Answer them and compare with what you thought then.

Check yourself: You're asked for "an appointments dashboard for the whole clinic." What do you do first?

Check yourself: Last week there were 402 scheduled appointments; 28 patients canceled with notice and 43 didn't show up or call. What is the no-show rate?

Check yourself: You give your data to an AI and it hands back a dashboard that says "No-show rate: 13.7%." It ran code to calculate it. What do you do before sharing it?

Reflect: Which dashboard or report at work will you redo first, and what decision should it support?

Only you and your Maestro read this. It's kept on your account and deleted with it.

Write your answer

What you can do now

Write down who a dashboard is for and what they decide with it.
Pick three to five numbers and define each one on a definition sheet.
Calculate rates and averages from totals, without averaging averages.
Give each number a comparison, in percentage points when it's a rate.
Use one chart per question, in the order of the decision.
Ask your AI for the dashboard with a brief, and check what comes back before you share it.

If you're not sure where to start with your own dashboard, talk it through with your Maestro.

If you want to keep going

This course stops at asking and checking, one conversation at a time. With the brief and the list you can already build your next dashboard; you don't need anything else to use them.

If you make charts with your AI often and want it to follow your taste without reminding it every time, the next step is the paid course Data Viz Co-Pilot. There you learn to read a chart with a designer's eye and fix it, you write the style guide and the skills your agent works from, and you finish by delivering a real chart you can defend.

Paid course: Data Viz Co-Pilot