Weekly recap: The thing converting trials to paid right now isn't a feature I built

This week I stopped trying to improve the product and started focusing on what happens between someone installing Genie 007 and them actually using it in a real work situation for the first time.

The data was uncomfortable.

Of the 22 trial starts last week, a solid chunk made it to their first real dictation session. The rest installed, opened the app, and went quiet. Not uninstalled. Just dormant.

I pulled up the drop-off points. Every silent account had stopped at the same place: the mic permissions step. Not because anything was broken. Because I had never explained why the permission was needed. There was just a button. No context.

Fixed it Wednesday. Added one sentence: "Genie needs your mic to convert speech to text in real time. Nothing is recorded or stored." 14 words.

Trial-to-first-session rate jumped roughly 20% in three days.

Not a product change. A copy change.

There's a version of this I recognise from sales. Deals used to stall for reasons that had nothing to do with the product. Someone couldn't get internal sign-off. They didn't know how to pitch it to their manager. My job was to find that friction and clear it. The product was only ever part of the story.

I'm applying the same lens to Genie 007 now. The product isn't the bottleneck anymore. The path through the product is.

Other numbers from this week:

• Chrome Web Store rating: 4.4 stars
• Main acquisition this week: word of mouth from one LinkedIn post about typing fatigue that drove more installs in 48 hours than a full month of Google Ads
• Ads: turned them off. Not a difficult call.

The LinkedIn thing keeps proving itself. One honest post about a real problem does more than paid traffic with a fraction of the cost. I'd rather understand that distribution advantage than keep paying to ignore it.

Next focus: the onboarding email sequence. Right now it's one generic welcome email. Building it into a 5-part flow based on what actions people have or haven't taken in the first 48 hours.

If you've done activation work on a solo SaaS: what was the smallest change that had the biggest impact on trial-to-paid? Genuinely curious what's worked for people at this stage.