Field test · Plot. · 10 remote sessions

Plot.

I made a chart styling tool for my own case studies. Then I watched 10 people make a chart with it.

My roleDesigner, moderator, and researcher. I directed the build with Claude Code.
Testers10 people, ages 19 to 36: students, marketing, finance, a data analyst, a researcher, and UX designers
MethodRemote moderated think-aloud on Google Meet, 10 tasks, about 40 minutes each
WhenNovember 2026
Why test it

I know where every button is. That is the problem.

I made the research charts in my case studies with Plot., so I never get lost in it. I wanted to see what happens when someone opens it cold, with real data and a real deadline in mind.

I recruited a mix on purpose: people who chart every day, people who design every day, and people who mostly use Canva or Google Slides. I treated their jobs as context, not as a guess about their skills.

The test

One dataset, one deadline, 10 tasks.

The scenario: "You are preparing a small report and need a chart that tells a simple story. Make it presentation-ready and export it for another design tool." Everyone used the same data, five months of website visitors with a small dip in April.

  1. Pick a chart type for comparing values.
  2. Enter the data.
  3. Add a title and labels.
  4. Change the colors so it reads well.
  5. Adjust the text to match a report.
  6. Change a chart setting like the axes, legend, or grid.
  7. Make the background transparent for a colored slide.
  8. Review it and fix anything that looks off.
  9. Export it as PNG or SVG and say why.
  10. Start over or undo without losing work.

I used neutral prompts like "What do you expect this to do?" and never pointed to a button. If someone needed a hint, the task counted as partial.

Results

92% of tasks done. Styling was easy. Starting was not.

92%of tasks completed (92 of 100)
10 of 10finished at least 8 tasks (4 finished all 10)
4.2 / 5average satisfaction

People out of 10 who completed each task on their own.

Task work took about 25 minutes on average. The data analyst finished in 19. The two slowest were undergrads who had only made charts in Google Sheets and Canva, at 31 and 32 minutes.

What I learned

The first and middle steps had the friction.

Styling is why people are here.

Everyone changed the colors and titles without help. One undergrad put it best: "I can make the chart look nicer than a default spreadsheet chart."

Starting was the hard part.

Two people pasted data in a format Plot. did not expect, and some picked the prettiest chart instead of the clearest one. "Once I found the data format, it made sense. A sample chart would help me start."

Text and chart settings were hiding.

People were not sure which control changed the title, the axes, or all the text. Controls spread across panels made that worse.

Everyone wanted a safety net.

All 10 found a way to start over, but most went looking for undo or reset first because they were scared to lose their chart.

It is a finishing tool, not a spreadsheet.

Data and finance folks loved it for polish but will keep doing the math in Excel. That is fine. It tells me what Plot. should not try to be.

SVG made sense to designers only.

Everyone exported, but non-designers did not know when transparency was kept or why SVG helps in Figma.

In their words

Real quotes from the sessions.

What I am changing

Fix the start, group the controls, add undo.

Now
→

Chart types are icons only. The data box is empty, with no example.

Next
Bar: compare amounts Line: show change over time Pie: show parts of a whole

Plain-language hints on every chart type, and sample data you can paste over.

P0

Chart-type guidance

A short plain-language line and preview for each chart type.

P0

Sample data and paste format

Show an example dataset so people know what shape their data should be.

P0

Group the styling controls

Put text, axes, legend, and grid controls under clear labels with live preview.

P1

Undo and reset

Visible undo, plus reset styling, so experimenting feels safe.

P1

Export helper

One line explaining PNG vs SVG, with a preview.

P1

Brand presets

Save a palette and reuse it, without locking designers in.

P2

Accessibility checks

Warn about low contrast and hard-to-read labels.

Honest notes

What this test cannot tell me.

Next round: ship the P0 fixes, then retest with 6 people (some new, some returning), with extra focus on students and marketing folks who struggled most with starting.

Next field testStick it!See how 10 people built a board →