April 27, 2024
Today was one of those “sweat the details or watch the ship leak” kind of days.
🔧 Bug Hunt Mode: Engaged
We’re neck-deep in stability work. That damn Safari bug crept up again—some sort of weird CSS rendering issue that only shows up when the moon is full and the stars align (or when our earliest users insist on testing in Safari 13). Spent way too long debugging weird z-index behavior that turns our onboarding modal into a ghost.
Speaking of onboarding, we had conversion drop-offs at the first step. Not great when you're trying to get feedback. Turns out a race condition was breaking things in 7% of sessions. That’s a conversion-killer. Fixed now. Fingers crossed.
📊 Spying on Our Users (for Good)
Event tracking finally started to take shape. Decided to give Mixpanel a go for the first implementation. It’s not fancy yet—we’re just tracking:
- Sign-up
- Onboarding complete
- First successful session
Noticed how easy it is to overtrack and underanalyze. So, took a minimalist approach: only track what we’ll actually act on.
🧠 Pitch, but Make It Real
Updated the deck. Tightened up the visuals, clarified the story. No fluff. No “we’re going to change the world” nonsense. Just: here’s the problem, here’s how we’re solving it, and here’s proof people care.
Had one warm intro turn lukewarm today. Not dead—it’s a slow dance. Still, every time I polish the deck, I get a little clearer on what we’re doing. Kind of cathartic, actually.
☕ Team Rhythm
Did a quick 15 with the team mid-day. Nothing fancy—just synced, shared progress, and joked about which browser is most evil (Internet Explorer is dead but its ghost possesses Safari now). Morale matters. Anxiety creeps in faster than you expect during these sprints.
💡 What I Learned Today
How to Choose What to Track (Without Hiring a Data Team)
If you're pre-product-market-fit, here's the rule of thumb:
“Track only the actions that tell you if users are succeeding.”
The goal isn’t to build charts. It's to learn. Here's how I approached it:
- Write down your product's 3 most critical user flows.
- Choose the success event in each (e.g., setup complete, message sent, first use).
- Ignore vanity metrics. If it doesn't answer "are we fixing the right thing?"—drop it.
- Use names you'll understand in 30 days. “onboarding_step1_complete” > “event_42”.
You don’t need dashboards yet. You need patterns. That’s enough to make better bets tomorrow.
Stable product, real signals, human pace. Felt the foundation firming up today.
Still lots to do. But it's finally starting to feel like a thing.
—Richard
