I error on the side of keeping other peoples information private. For the tools on my website, this meant simply putting them out there. However, in order to know where I should focus my time I need to know if anyone is even using the tools. I also need to provide a mechanism for contacting me about issues or improvements.
Yesterday I added telemetry beacons to every tools page on my site. Color picker, handwriting font matcher, star map, QR generator, all of them. Every page load fires a beacon to an Apps Script endpoint that writes a row to a sheet. I’ll have a count by tool, a timestamp, a referrer if the browser sends one. That number is the ceiling: everyone who landed and tried something. I only collect the usage information, not the user information. For example, the starmap information collected includes if the user started with a preset and which style settings were selected. I DO NOT save the date, location, message or the final image.
Then I added a feedback form.
Not because the telemetry felt wrong, but because it answers the wrong question. A row going into my sheet every time someone loads the color picker tells me someone was there. It has nothing to say about whether they pointed their phone at a paint swatch, got back “Cornflower Blue,” and left more confused than when they arrived. Or whether they found Simple mode (12 everyday names instead of 148 CSS names) and that was exactly what they came for. The visit is recorded either way. The outcome isn’t.
The feedback form asks two things: did the tool do what you needed, and what would you change. Kudos has its own version, but Kudos is an appreciation platform. I built it so people can tell each other something good. That’s a different use case from “the QR generator exported at 72 DPI and I needed 300.” I had infrastructure for gratitude. Honest critique needed its own lane.
Here’s what I’m expecting: the telemetry number will be a lot bigger than the feedback number. That’s probably fine. Most of the gap represents people who used the tool, got something out of it, and moved on without any reason to report back. But some of that gap is people who tried it, didn’t find what they needed, and didn’t bother saying so. I can’t tell those two groups apart from a row count.
And you know what, that’s true of most things anyone ships publicly. The telemetry ceiling and the feedback floor are both real. What matters is how the distance between them changes as the tools improve.
If you’re building something for people you’ll never meet, ship the “did this help” channel before you spend more time on features. Not because you’ll get a flood of responses. Because the few you do get are worth more than the row count will ever tell you.