FlowProofsData storytelling is a phrase I avoided for years, because it sounded like something you would hear at a conference right before someone showed an animated pie chart. But the problem it names is real: you have the right numbers, and the meeting still goes badly because nobody can follow what they mean.
So here is the version without the buzzwords. Eight practical rules for presenting numbers to people who will question them, with one running example.
The example
A distribution business. Gross margin fell from 34% to 31% this month on flat revenue of $1.33M. The cause, after a day of digging, is freight: shipping costs rose $40K, mostly on one carrier's new fuel surcharge.
That is the whole finding. Watch how easily it gets lost.
The typical deck opens with a revenue chart (flat, so nothing to see), then a margin trend over twelve months, then a cost breakdown table with fourteen lines, then a carrier comparison, and finally, on slide nine, the sentence about the fuel surcharge. By slide nine, the finance director has already asked about payroll, and the room is now discussing payroll.
Rule 1: one question per view
Every chart, table or slide should answer one question. Write that question down before you build it. If you can't phrase it, the view doesn't belong in the meeting.
For our example, the questions are:
- Did margin fall? (Yes, 34% to 31%.)
- What drove it? (Freight, +$40K.)
- Where is the evidence? (The freight file, 212 shipments.)
Three questions, three views. Not fourteen cost lines.
Rule 2: put the answer in the title
Most slide titles are labels: "Gross margin trend", "Cost breakdown". A label tells the reader what kind of chart it is. They can see that already.
Write the conclusion instead: "Margin fell 3 points because freight rose $40K." Now a reader who only skims the titles still leaves with the finding. Consultants call these action titles. Whatever you call them, they force you to decide what the view means before you show it.
Rule 3: put the driver next to the total
A total moving is a question. The driver is the answer.
When margin drops, the margin number on its own invites guessing. Was it price? Mix? Costs? Put the bridge next to it: margin last month, each cost line that moved, margin this month. The room sees immediately that freight is the only bar of any size.
This is the same idea as decomposing revenue into drivers, which I walked through in how to explain why revenue changed. The arithmetic is simple. Doing it before the meeting is the hard part.
Rule 4: one comparison, chosen on purpose
Every number needs something to compare it to. Pick one baseline per view and say which: last month, the same month last year, or budget.
Showing all three at once feels thorough and reads as noise. If the question is "is this unusual?", compare to last year. If it is "are we on track?", compare to budget. If it is "did something just change?", compare to last month.
Rule 5: show the source
The fastest way to lose a room is "where does that number come from?" followed by a pause.
Put the source on the page, small but visible: "Freight costs from the carrier invoices file, 212 shipments, September." It costs one line and removes a whole category of doubt. It also protects you. If the source turns out to be wrong, everyone can see where the problem started.
Rule 6: say what didn't change
This one is underrated. When you present the driver, also say what you ruled out: "Revenue was flat. Product mix was stable. Payroll was on budget."
Each of those sentences closes off a line of questioning before it starts. Without them, the finance director's payroll question is a reasonable thing to ask, and it costs you ten minutes.
Rule 7: label directly, drop the legend
Legends make the reader look back and forth between the chart and a key. Put labels on the lines and bars themselves. Write the actual values on the bars that matter. Grey out everything else.
A chart where one bar is coloured and the rest are grey tells the reader where to look before they have read a word.
Rule 8: know what you want the room to do
Before the meeting, finish this sentence: "By the end, I want them to decide ___."
For our example: decide whether to renegotiate with the carrier or move volume to the second carrier. That decision is the reason the meeting exists. If your views don't lead to it, people will fill the time with the questions they find most interesting, which are rarely the ones that need deciding.
The same finding, presented twice
Here is the margin story laid out both ways, so the difference is concrete.
The nine-slide version. Slide titles: "Revenue overview", "Gross margin trend", "Cost of sales breakdown", "Freight by carrier", "Freight by region", "Payroll", "Packaging costs", "Carrier contract terms", "Summary". The finding appears on slide nine, as a bullet. Every slide is accurate. The meeting runs forty minutes and ends with three follow-up actions, none of them about the carrier.
The three-view version.
- "Margin fell 3 points to 31% on flat revenue of $1.33M." One number, last month beside it.
- "Freight explains all of it: +$40K, mostly one carrier's new fuel surcharge." A bridge from 34% to 31%, with freight as the only large bar and a line underneath saying payroll, packaging and mix were flat.
- "Decision: renegotiate with Carrier A, or move volume to Carrier B." Both options with their cost, and the source file named at the bottom.
Same data and the same analysis behind both. The second version takes about ten minutes and ends with a decision. The first version still exists. It just moved to the appendix, where the people who want it can find it.
When these rules work against you
A few honest caveats.
Some audiences want the detail. An analyst reviewing your method, or an auditor, needs the full cost table. Give it to them, in an appendix or a follow-up. The rules above are for the decision meeting itself.
Storytelling can become spin. If you choose the one baseline that flatters your number, you are not simplifying. You are hiding. The test I use: would I be comfortable if someone added the comparison I left out? If not, put it back.
Not every finding is a clean story. Sometimes margin fell for four small reasons and none of them dominates. Say so. "No single driver: four lines each moved about $10K" is a perfectly good finding, and much better than forcing one narrative onto it.
Applying it to a dashboard
These rules came out of slides, but they apply just as much to dashboards, which tend to break most of them. A typical dashboard shows twelve tiles with label titles, no drivers, no sources and every possible comparison. It answers every question a little and no question fully.
The fix is the same: decide the questions first, put the answer in the title, and connect each headline number to the numbers that drive it. A metric tree is the most compact way I know to do that for a single number.
This is the habit FlowProofs is designed around. Numbers sit on a canvas next to the numbers that produce them, every value shows its source table, and the why panel calculates which input moved. Consultants who send clients reports use it so the client can click a number and see the driver themselves, which is the setup on the consultants page. But a well-built slide with the eight rules above gets you most of the way.
A one-page version
If you remember nothing else, before your next meeting:
- Write the questions the meeting needs answered. Cut any view that doesn't answer one.
- Rewrite each title as a conclusion.
- Next to every headline number, put the driver that moved it.
- Choose one comparison per view, and say which.
- Add the source in small type.
- Add one line on what didn't change.
- Label directly. Delete the legend.
- Write down the decision you want, and end on it.
Then present the three-view version of your next finding and time how long it takes before the first question. If the first question is about the decision, it worked.