Launch day numbers

I want to be the person who shares real launch numbers instead of the person who talks vaguely about "strong traction." Here's the actual data from day one.

### The numbers

Downloads: 847

Units by source:

• App Store search: 312 (37%)
• Referral links (newsletter, social): 218 (26%)
• App Store Browse (featured/category browse): 156 (18%)
• Direct (typed the App Store link): 89 (11%)
• Unknown/organic: 72 (8%)

Sessions: 1,234 (some users opened multiple times)

Boards snapped: 423 (first-day activation: 50%)

Pro upgrades: 31 (3.7% day-one upgrade rate)

Revenue: $279.69 (31 monthly subs)

Reviews submitted by end of day: 7 (all positive; early reviewers are obviously fans)

Crashes: 3 unique crash reports, all on the same iPad edge case (VisionKit behaves differently on iPad; I hadn't fully tested iPad targets)

### What I got right

App Store search was the top channel. I'd invested heavily in App Store metadata — keyword research, a/b tested subtitle, screenshots that showed the actual output rather than lifestyle photography. The payoff: 37% of day-one downloads came from people searching for things like "whiteboard scanner AI" or "meeting notes app."

The upgrade rate was meaningful. 3.7% on day one is respectable. These are early adopters who saw the value quickly enough to pay on the first day. Early upgrade rates often correlate with overall product-market fit.

### What I got wrong

I overestimated social referrals. I'd sent the launch announcement to my newsletter (~3,200 subscribers), posted on X, posted on LinkedIn. I expected 35–40% of downloads from these channels. Actual: 26%.

The newsletter had a 42% open rate and a 6.7% click-to-install conversion. Good, but less than I'd modeled. The X post had high impressions and low conversions. LinkedIn drove more installs per impression than X, which was the opposite of what I expected.

The iPad crash was embarrassing. I'd tested on iPhone extensively. I'd tested on the simulator for iPad. I hadn't tested on a physical iPad with the camera scanner flow. There's a specific behavior difference in how VisionKit handles landscape orientation on iPad that caused a crash on a capture attempt. Three users hit it within 3 hours. I fixed and resubmitted within 90 minutes of the first crash report.

The lesson: test on physical devices for every target form factor, not just the primary one.

### The day-two perspective

Day one numbers are the wrong thing to optimize for. Day seven retention, Day thirty revenue, word-of-mouth that shows up in week three — these are the metrics that actually matter.

But day-one numbers tell you if you've got signal or static. 847 downloads, 50% activation, 3.7% day-one upgrade: that's signal. The product is landing for at least some of the people who find it.

I'll write a thirty-day update when I have it.