The Slide Made Perfect Sense to Me: Why Every Chart Needs a Test Audience
A chart you built yourself will always make sense to you. That’s exactly the problem. You already know what the two lines mean, which number is good news, and where to look first; none of that is on the slide- it’s in your head. The only way to find out what’s missing is to hand it to someone without your head start. If someone has to ask what a chart means, the chart has already lost.
Many times, it is caused by your inner circle. Why? Well, because if you work with someone for a long time, your business partners, your team, your audience, you share the same common ground or baseline of business understanding. You use the same language among yourselves; you don’t have to describe things as much, because you have a lot of history in how you run the analytics, dashboards, and outcomes. And we think that if this group of tightly connected people understands each other, everyone will.
Well, that’s a big misunderstanding. If the insights do not travel without you, the mission has failed. You can’t be in every meeting or call. Your analysis, if it’s good, doesn’t always stay within your circle. The new audience doesn’t understand the shortcuts, abbreviations, and language you use. You need to tailor your message to the right audience. But you can only find out by testing it with someone first.
If you remember nothing else from this tip, take a little time to get feedback from peers, other managers, or leaders before your next big review. It will pay off many times over, since you usually don’t get many chances with leadership.
This is from the series of TOP 30 Tips in Data Storytelling.
“Without feedback, effort turns into repetition instead of improvement.”
The Curse of Knowledge Is Working Against You Right Now
In 1990, Stanford psychology student Elizabeth Newton ran a study that still gets cited whenever someone tries to explain why smart people build confusing slides. She split volunteers into two roles: “tappers” and “listeners.” A tapper picked a well-known song, “Happy Birthday” or “The Star-Spangled Banner,” and tapped its rhythm on a table with one finger. The listener’s only job was to name the song.
Before tapping, the tappers guessed listeners would get it right about half the time. The real number was 2.5%. Out of 120 songs tapped, listeners named three.
Here’s the part that should make you uncomfortable: while a tapper’s finger hit the table, they could hear the entire song in their head – melody, lyrics, the works. They genuinely couldn’t imagine what it sounded like from the other side, where all that arrived was a knuckle knocking on wood.
Chip and Dan Heath later built an entire chapter of Made to Stick around this finding and gave it a name: the curse of knowledge.
Once you know something, you lose the ability to imagine not knowing it.
Your chart is the tapping. The six hours you spent staring at that dataset is the melody only you can hear. Your audience gets the knuckle on the table and then takes the blame for not “getting it.”
Sometimes it is that you don’t know your AUDIENCE or your WHY – I go deeper into this here in TIP 1 and TIP 2.
Example: Marketing
Rough Situation: An 8-week campaign dashboard tracks weekly spend and conversion rate for the CMO’s Friday review. Spend rises steadily. Conversion rate dips sharply in week 6, then recovers in week 7.
Situation:
Spend and conversion rate genuinely belong on the same chart. They’re two views of the same campaign, and viewing them together over the same time periods is exactly the point of the review. Different scales aren’t the problem here. The problem is that the week 6 dip has a real, specific cause: a three-day analytics tracking outage that undercounted conversions, with nothing to do with the campaign itself, and the chart doesn’t say so.
How not to do it:
Show spend climbing steadily right through the week when conversion drops, with no note about why. Send it to the CMO as-is. A reasonable person looks at spend rising at the exact moment conversion falls and draws the obvious conclusion: heavier spend is buying worse traffic and dragging performance down. Nobody said that. The chart simply left a blank space, and a plausible-sounding story filled it in on its own.
How to do it:
Show the same draft to one colleague first, cold, with no context. If their first reaction is “wait, so more spend is hurting conversion?”, that’s the test working exactly as intended. They didn’t misread the chart; they read it correctly, and it told them the wrong thing. Add the real explanation directly on the chart, right at week 6: a short note naming the tracking outage. The invented causal story disappears the moment the true one is sitting in front of it.
I learned this the expensive way. Early in my career, I built what I was sure was an airtight slide for a steering committee: three years of regional sales on one chart, color-coded by product line, with a dotted trend line cutting through the middle to show where things were headed. I’d spent two full days on that slide. I could have explained every pixel of it in my sleep.
Thirty seconds into presenting it, the senior leader asked, “Sorry, which color is this year?” Someone else asked what the dotted line was even measuring. By the time I finished answering both questions, we’d run out of time to discuss what the chart was actually built to prove.
Nothing on that slide was factually wrong. It was just built entirely from inside my own head, and nobody in that room had spent two days living inside it with me the way I had. I changed one habit after that meeting, and I’ve kept it ever since: every chart gets shown to one colleague, cold, before it goes near a room I can’t fully control.
It’s the cheapest insurance I’ve found in this job, and it has saved me from that exact feeling more times than I’d like to admit.
All of the core principles are described in my simple training for anyone: The Curious Beginner.
Let’s move to a second example, where the problem wasn’t the chart type, it was everything competing for attention inside it.
Example: Operations
Rough Situation: A plant-efficiency dashboard ranks 12 production lines by on-time delivery percentage, built to tell floor managers which line needs support this week.
Situation:
The dashboard exists to answer one question: which line should get support this week? The Before chart answers a different one: which line has the worst on-time rate, and it treats the two as the same question. They aren’t. Line 7 has the worst rate on the floor, but it barely ships anything, so a bad rate there still means only a trickle of late orders. Line 3 has a respectable rate, but it ships far more than any other line, so that same small percentage gap adds up to more late shipments in absolute terms than anywhere else in the plant.
How not to do it:
Rank all twelve lines by on-time percentage and send the chart straight to the floor managers. Line 7 sits at the bottom with 68%, and every eye goes there immediately; it’s the obvious read, and it isn’t wrong on its own terms. Someone gets sent to fix Line 7. Line 3, generating more late shipments than any other line in the plant, keeps running exactly as it was, because the chart never asked the question that would have surfaced it.
How to do it:
Show the ranked-percentage chart to one colleague, ideally someone with no stake in the plant, and ask what they’d do with it.
“I’d get someone on Line 7, obviously. It’s dead last.”
“Fair. How much does Line 7 actually ship in a day?”
“…I don’t know. Does it matter?”
“It ships 40 units. Line 3 ships 900.”
“Oh, so which one has more late shipments, not which one has the worse rate?”
That last question is the one the original chart never asked. It’s not a rephrasing of “which line has the worst rate”; it’s the actual business question the dashboard was built to answer, and it only surfaced because someone outside the situation said the obvious answer out loud and got asked to defend it. Rebuild the chart around that question: late shipments per day, not percentage, and move Line 3, not Line 7, to the top.
The two failures above don’t look alike on paper. Marketing hid a missing explanation. Operations hid a missing variable. What ties them together sits off the page: both charts survived every internal review, and neither survived five minutes with someone who didn’t already know the intended answer.
That’s the part worth carrying forward, more than either example. The specific gap will be different next time: a missing cause, a missing variable, something else entirely.
The fix stays the same: one outside reader, one honest first reaction, before the chart reaches anyone with real stakes in it.
How to Test a Story Without Turning It Into a Committee Project
Testing doesn’t mean a design review, a survey, or a week of waiting on feedback. It means one person, one cold read, before the chart leaves your laptop.
Three checks worth running before any chart ships:
- Hand the chart to someone unfamiliar with the topic and ask them to say out loud what it means. If their answer doesn’t match your headline, the chart needs work, not the person.
- Time how long it takes them to find the one thing you built the chart to show. A few seconds is a pass. A minute of scanning is a chart that needs sorting, highlighting, or fewer elements.
- Ask what they’d do next after seeing it. If the honest answer is “nothing,” the chart hasn’t earned an action yet, and neither has the story around it.
None of this requires a formal process. It requires being willing to hear that the chart you were proud of five minutes ago didn’t land the way you thought it would, and fixing it before anyone with real stakes sees the first version.
Summary – From insights to action
A chart that only makes sense to the person who built it isn’t finished, no matter how long they spent on it. The fastest, cheapest edit available to you is someone else’s honest first reaction, and it costs nothing but ten minutes and a little bit of ego.
Before you send your next dashboard, deck, or one-pager, find one colleague who has no context for it and watch them read it cold. Whatever confuses them will confuse your real audience too, just later and with more at stake.
Don’t forget to try the free 10 Step Training for Curious Beginner here.
For free resources to help you on your Data Storytelling Journey, click here.
To keep Transforming Your Data Storytelling Journey, sign up for regular Insights.
Leave a Reply