Built what you told me to build
A couple weeks ago I posted an honest recap here — App Store listing fixes, some SEO content, zero installs moved. The comments turned into one of the more useful things that's happened since launching: several of you, here and on the Product Launch thread, kept pointing at the same gap — the product was just automating a habit (checking numbers), not actually telling merchants anything useful. "Insurance against a bad week," not a weekly export.
Didn't want to just agree and move on, so here's what's shipped since:
- A predictive alert engine (6 rules: revenue drop, order drop, AOV drop, anomalies, no-sales periods)
- A recommendations engine — actionable suggestions, never an invented cause
- A natural-language insights layer explaining what's actually happening, not just numbers
- Configurable alert thresholds
- Excel export alongside PDF/CSV
- Rebuilt the whole PDF/email pipeline on one shared data source, so the two can't drift out of sync again
- 186 automated tests across all of it
Also went back and fixed the pricing page, which — until someone here pointed it out — still just said "unlimited reports, email delivery, email support" while the product had already grown past that.
Honest state: no idea yet if any of this moves installs or reviews. Still close to zero on both. But the product is a genuinely different thing than it was three weeks ago, and that happened because of comments in this community, not because I sat and brainstormed alone.
If you were one of the people who pushed on this — or weren't, and have thoughts anyway — curious what you'd want to see next: more alert types, copy that leans harder into the "insurance" framing, or something else entirely.
You’ve now changed the product from reporting to decision support, but the near-zero installs make the next test pretty important. What evidence would tell you merchants actually want the “insurance” outcome before adding more alert types or polishing the messaging?
Fair challenge. With install numbers this small, I can't wait for statistically meaningful data — so the evidence I'm actually planning to look at is behavioral, not volume-based: does an alert email get opened/clicked at a noticeably different rate than the routine weekly report, for the same person? That's a within-user signal, not a sample-size one — even a handful of merchants can show it if the pattern is real.
Beyond that, the more honest answer is I don't have a clean test running yet. I've been talking about instrumentation more than shipping it. Given where things actually stand, the more useful next move is probably closing that gap — tracking open/click per email type — rather than adding alert types or rewriting copy based on a hypothesis I haven't checked.
What would you consider strong enough evidence, with a user base this small?
That behavioral test is the right direction. If you’re open to it, what’s the best email to reach you on?