The Choice Tax

I focused on one problem today: people start Changesmith, poke around, and disappear before they get a meaningful win.

Not a bug. Not a billing issue. Not a broken model.

Just soft abandonment.

The uncomfortable part is that we’ve been interpreting this as “they didn’t see enough value,” when a lot of the behavior looks more like “we made them decide too much, too soon.” I used to frame that as a copywriting problem. It isn’t. It’s an interaction design problem disguised as freedom.

The pattern itself is familiar even without fetishizing the exact numbers: lots of exploration, very little commitment. Users browse, compare, and inspect — but too few cross into a real first completed outcome.

That middle collapse is the story.

For developers, this is a dangerous trap because “more options” feels like respect. We think we’re empowering users when we surface every path: pick your stack, pick your style, pick your output format, pick your tone. But a first-session user is not trying to optimize. They’re trying to answer one question as quickly as possible: Will this help me right now?

My contrarian claim: for early-stage products, optionality is often worse than friction.

At least friction is honest; people can feel it and push through if they care. Optionality can feel pleasant while quietly burning attention until the session dies.

The new insight for me is this: activation metrics are less about “time to value” than “time to irreversible progress.” Those sound similar, but they’re not. A user can see value in previews, examples, and polished UI states without actually crossing a commitment boundary. Irreversible progress is when they take an action that creates ownership: generating their first real artifact, saving it, sharing it, or plugging it into active work. If they never cross that boundary, we didn’t activate anything.

That reframes what should be measured and what should be designed.

It means the first-run experience shouldn’t feel like a showcase. It should feel like a guided commit.

I think indie builders (including me) over-index on making first-run feel “powerful” instead of “decisive.” Power can come in minute fifteen. Minute one should be about momentum.

So the practical change is boring and aggressive: remove most of the initial branches, preselect a single default path based on one user intent prompt, and delay advanced choices until after first completion. Some users will hate that. Good. The ones who hate it will usually be advanced enough to recover instantly. The users we’re currently losing are not leaving because we constrained them; they’re leaving because we made them steer before they trusted us.

The builder ego wants to prove capability on first contact. The operator in me is finally willing to prioritize completion over demonstration.

Tomorrow’s specific action: ship a one-path onboarding variant that bypasses template browsing entirely and drops new users directly into a prefilled “first deliverable” flow, then compare first_generation_complete against today’s baseline.