The ten-second rule

I arrived at the ten-second rule empirically, not theoretically.

I watched beta users at the moment a meeting ended. The meeting is over. People are starting to move. There's a window — maybe 90 seconds — where the whiteboard is still on the wall and the room still belongs to you. What tool do you reach for in that window?

I timed it. Users who reached for their camera app (no BoardSnap) took an average of 4.2 seconds from the end of the meeting to the photo. Users who opened BoardSnap took an average of 8.7 seconds.

The 4.5-second gap was the competition. When the meeting ends, the camera app is already there. You open it, point, shoot. Done. If BoardSnap took 15 seconds, users wouldn't bother — the camera app is faster and good enough for the job they're doing (taking a photo, posting to Slack).

At 8.7 seconds, the additional value BoardSnap provides — the analysis, the action items, the structured output — justifies the additional 4.5 seconds over taking a plain photo.

But that justification has a ceiling. I modeled it: if BoardSnap's core flow took 15 seconds, the majority of users would default to the camera app. If it took 5 seconds, it would be genuinely competitive with the camera app even for users who weren't thinking about action items.

Target: under 10 seconds, camera open to snap confirmed.

### How we built to the threshold

The cold start problem (covered in a separate post) was part of this. A 3.2-second startup ate most of the budget before the user even opened the camera.

But the in-flow steps matter too:

• Camera to detection: VisionKit runs in real-time in the viewfinder. By the time you've pointed the phone at the board, detection is usually already locked. Zero marginal time.
• Detection to capture: one tap on the shutter. We resisted the temptation to add a confirmation dialog at this step. One tap, committed.
• Capture to analysis: the image is immediately queued; the capture confirmation card appears in under a second. Analysis happens in the background. Total blocking time: under 1 second.

The blocking portion of the flow — camera tap to confirmed capture — is under 3 seconds on a warmed app. The analysis is non-blocking. You can leave the room while it runs.

### Why ten seconds specifically

The ten-second threshold isn't magic. It's approximately the time it takes to open a fresh app on an iPhone, navigate to a capture function, and take a photo. If BoardSnap's blocking flow is within that budget, the app doesn't feel slower than the alternative.

There's also a meeting-transition psychology to it. When a meeting ends, you have about 90 seconds of focused attention before the transition to the next thing fully absorbs you. Ten seconds is comfortably within that window. Thirty seconds is not.

### The headline number

The BoardSnap homepage says "ten seconds." That's the number that tests well, it's the number that's accurate on warm start, and it's the number that matters for the meeting-end use case.

On cold start (first-ever launch or after a week with no opens), we're closer to 12–14 seconds to the first completed snap. On a warmed app — which is almost always the case for a regular user — we're under 10.

The ten-second target is a constraint that runs through every performance decision in the app. When something makes the critical path longer, we notice. When we ship a fix that brings it back, we celebrate.