Case Study 04 · Product Strategy, Design & Build
Increasing marketing visibility for an open source platform
Shipped features for Excalidraw, an open source whiteboard tool. The marketing KPI: grow the share of exports that carry visible attribution, to increase the platform's visibility and support growth. Worked as UX researcher and product designer inside a shared GitHub repo with three other contributors. Led product design and ideation on subtle UX changes including QR codes, default credit line, and a short opt-out survey. Also contributed UX edits to the export-tracking dashboard built by team to measure the KPI.
01 · The Brief
The stakeholder was Excalidraw's Head of Marketing, played by a Pursuit staff member. Their KPI was brand visibility rate: the share of exported or shared drawings that carry visible Excalidraw attribution.
Two constraints shaped the work. First, Excalidraw is local-first. There are no accounts, and no server ever sees a drawing. Any idea that needed a login, a database or a tracked user was ruled out. That is why the plan relies on small, transparent, optional UX changes.
Second, there was no baseline. Existing analytics couldn't tell whether an export carried attribution, so there was no clean export rate. Rebuilding the export dashboard around this KPI became part of the deliverables.
02 · Research
Most Excalidraw drawings do not leave through a "File > Export" menu. They leave as a copy-pasted PNG, dropped straight into a GitHub pull request, a README, a blog post, or a slide. Any attribution that only lived inside an export dialog, or behind an opt-in toggle, would miss most of the real usage.
| Tool | Free-tier export attribution |
|---|---|
| Miro, Whimsical, Napkin.ai | Attribute on the free tier |
| tldraw.com, FigJam, draw.io | No watermark on export |
| Canva | Watermarks only when premium elements are used |
| Loom | Badge on nearly every free video |
Loom puts its badge on nearly every free video automatically, which is why it works. But Loom is a private, commercial product, and it later charged people to remove the badge. Users called that a "tax." Excalidraw is open source and free, so that would go down even worse here. My rule for this project: the credit is on by default, but turning it off always stays free.
Developer
Lola is a developer, who wants to drop a clean diagram into a pull request or README in seconds, so that reviewers get context without a screenshot tool or extra editing. She has used Excalidraw for years and likes its no-pressure, local-first approach. If that changed, she would switch quickly.
Teacher
Mr. Smith is a teacher with no tech background, who wants to export a drawing for a class slide in one step, and turn off anything he doesn't understand without getting stuck. He doesn't use Excalidraw yet and hears about FigJam and Miro instead. He would try it if it were easy for non-technical people to navigate, and if he could export in different formats and aspect ratios for slides.
Student
Camila is a student preparing for system design interviews, who wants to share diagrams with her study group and on LinkedIn, so that others see her thinking. She is happy to credit Excalidraw, and her friends would rather back an open source tool than a private one. But she can't tell how to export at a set aspect ratio, and her homework needs to look consistent.
The direction we chose was to capture more visibility from Excalidraw's existing, engaged developer base, rather than build new features to chase a new segment like classrooms. Formal use in education was ruled out early. It needs procurement approval, stricter privacy handling, and classroom features Excalidraw does not have. Putting the same effort into the audience already producing shareable drawings every day was the higher-leverage bet.
Teachers and students who do find their way here still matter, whether they are tech students or non-technical tutors. So we made exporting easier for them, with more formats and aspect ratios. That gives them more ways to export and more reasons to keep using the tool.
03 · Design
The plan favored small, quiet signals that add up, rather than one loud watermark, like a credit line in the export margin by default, a QR code on shared collaboration links, and a survey that only appears if someone turns attribution off. Below is how they were ranked.
| Priority | Features |
|---|---|
| P0Core to the KPI |
|
| P1Supports the KPI |
|
| P2Next iteration |
|
04 · Building It
The implementation was coded with Claude Code, while the team directed and oversaw the product decisions, the design review, and the QA. Each result was checked on a local dev server before it was committed.
The scope and the trade-offs were fixed before any code was written.
Bug
The mark rendered in the middle of the image, and too small, instead of the bottom-right corner.
Cause and fix
The canvas was already scaled by the renderer, and the new mark code scaled it a second time. Fixed by drawing inside the context that was already scaled.
Bug
The mark got cropped off on small or tightly zoomed exports.
Cause and fix
The margin reserved for the mark was a guessed fixed size, too small for the mark it needed to hold. Fixed by computing the margin from the mark's real size, with a clamp so it can never land off-canvas.
Testing has a real limit since the test environment cannot rasterize a canvas, so pixel position bugs like these cannot be caught by an automated test alone. They were verified visually myself, using the SVG output, which can be inspected directly
05 · Teamwork
The work landed as three commits on one branch, in one pull request. Three teammates worked on their own branches in the same repository, so we kept each unit small and checked what others were building before starting the next one.
The commits are ordered by dependency: mark and opt-out first, then export formats, then survey reporting.
Each one was typechecked, linted and tested when it was made, so any commit can be reviewed alone or undone
with git revert without breaking the others.
| Contributor | Work |
|---|---|
| Khushali H. | Export attribution mark, opt-out survey, export formats, aspect ratios |
| Bo M. | QR code logo, branded export filenames, file-type icon branding |
| Richard T. | Export analytics dashboard |
| Vincent M. | Live-sync embed with attribution, deployment configuration |
06 · Results
What's built: a default attribution credit on PNG, JPEG, WebP, and SVG exports and on copy-paste, a hide-attribution toggle gated by the survey with a 30-day expiry, a redesigned export dialog with more formats, and an aspect ratio picker. Survey answers are reported as anonymous analytics events.
What is not proven yet: there was no attribution baseline before this work, so I cannot say the KPI moved, only that it can now be measured. The analytics dashboard built by team still runs on mock data, and is not yet connected. Cross-browser testing is yet to be reviewed, since the mark uses canvas features that can vary between browsers.