Week 11: 2 new paid users, one brutal onboarding insight, what's next for Genie 007
Quick Monday update. Keeping this honest.
What shipped this week:
Fixed an onboarding flow issue I suspect was killing activation. New Chrome users were hitting a permissions screen with zero explanation of why it was there. Added one line of context. Felt obviously wrong when I finally watched someone try to set it up live.
Also pushed a fix to the Windows client causing a rare crash on startup. One user reported it. Probably others just quietly uninstalled.
The numbers:
2 new paid users this week (both from organic search, not social)
Churned 1 user who'd been on the monthly plan for 6 weeks. No reason given.
Chrome extension installs: up 12% week on week
Free to paid conversion: sitting at 4%. Want to get this to 7-8%.
What I learned:
The churn stung. Not because of the £9/mo lost. Because I had no idea why. I don't have an offboarding survey. Building one this week, 3 questions max.
Also: 2 of my last 3 signups came from a Reddit thread I had nothing to do with. Someone recommended Genie 007 in an RSI/ergonomics forum. That one unprompted mention probably did more than 6 weeks of LinkedIn posting combined. I need to find where those conversations are happening and be part of them.
What's next:
Offboarding survey. Mention monitoring across Reddit and niche forums. A/B test on landing page hero copy ("talk to your computer" vs "type less, do more").
The honest bit:
I keep going back and forth between "this is working" and "is this actually going anywhere." Last week felt like the latter. This week feels more like the former. Probably just noise.
Anyone else find the weeks you feel most stuck tend to be followed by something clicking? Or is that just a story I tell myself to keep going?
Really appreciate how honest this is — especially the part about going back and forth between “this is working” and “is this going anywhere”.
I’ve been feeling that exact same thing building an AI UI tool.
One thing that stood out to me from your post:
that onboarding fix from actually watching a user.
I had a similar moment recently — something that felt “obvious” once you see it, but completely invisible when you’re just building in your own head.
Also your point about the Reddit mention is huge.
It feels like:
1 real mention in the right context
I’m starting to think distribution for dev tools is more about being present where the problem is already being discussed, not trying to create attention from scratch.
Curious — when you saw those Reddit signups, did you try to jump into those threads or just observe for now?
Observed for now. Didn't want to show up in those threads with a pitch before I understood what conversation was actually happening. Once I've mapped what problem they're framing it as, I'll jump in with something useful rather than just dropping a link.
Really appreciate you sharing these honest numbers, tracking churn and conversion like this is gold. I often help founders set up lightweight surveys, monitor forum mentions, and organize onboarding/offboarding flows so you can spot patterns without getting overwhelmed. Happy to share a template or workflow if you want; it’s saved a few teams' hours each week.
The offboarding survey is actually on the list for this week — something I've been putting off because it felt too manual to set up. Appreciate the nudge.
Honest and helpful breakdown. Getting initial traction is rough. You mentioned knowing where the Reddit threads are where users discuss your product or a problem adjacent - I am building a tool for exactly this called Wotchrware, it watches community boards 24/7 for relevant threads, and drafts custom replies in your own writing voice for each thread. Human in the loop stays, all drafts go to the user for editing and posting, no automatic agent posting that could be the start of a dystopian movie 😅
I especially liked the off boarding survey idea to capture the "why", churn isn't fun but leveraging for improvement helps it sting less.
Good luck with getting started!
That's a genuinely useful angle — monitoring where the problem is already being discussed rather than broadcasting into the void. The human-in-the-loop framing is smart too, especially as people get more suspicious of AI-generated replies. Is Wotchrware live yet or still building?
it's still in the building stage, refining the onboarding for the agent to learn the user's writing voice in as efficient a way as possible so the initial learning curve is a short as it can be.
week 11 with 2 paid users and an onboarding insight is real progress. most people quit by week 3. the brutal onboarding realization is usually the most valuable thing you learn early on — we found the same thing with our dev tools. our original onboarding was "sign up then figure it out." switching to "paste a URL and see results instantly, no signup needed" changed everything. the fewer steps between landing and value, the more people stick around. whats the onboarding insight that hit you?
The permissions screen thing. Chrome extension users were hitting a "allow permissions" screen right after install with zero explanation of why it existed or what it would do. I assumed it was obvious. It wasn't.
Watched one person try to set it up live on a call and they just... stopped there. Confused, not blocked. Completely fixable with one sentence of context. That one session was worth more than 6 weeks of looking at drop-off data.
Your "paste URL and see results" shift sounds like the same principle from a different angle. Both are about getting someone to value before they have to commit. The faster you can do that, the better. Did you see activation numbers move immediately after making that change?