Freemium

Freemium works by removing the purchase decision barrier for new users. Instead of convincing someone to pay before they have experienced the product, you give them enough to demonstrate value, then monetize the subset of users who hit the free tier's limits and decide the product is worth paying for. Conversion rates from free to paid typically range from 2-5% for consumer tools and up to 15-20% for B2B SaaS with high switching costs.

The model's economics depend entirely on the cost of serving free users. For AI tools that call external APIs on every use, free tier users directly cost money. For tools that run entirely in the browser, free tier users cost essentially nothing beyond bandwidth. That difference decides how generous a free tier can be: metered daily limits are the norm wherever every request has a marginal inference cost.

The design question for freemium is where to draw the free/paid line. Too restrictive and users do not get enough value to see why they should pay. Too generous and users have no reason to upgrade. Coda One's model — 3 free uses a day for the Humanizer, Rewriter and Grammar Checker, with the AI Detector unmetered — reflects that cost difference. The detector is not cheaper because it runs locally; it does not. It is cheaper because it makes no model call: the text is scored by arithmetic over word and sentence statistics on a server that then forgets it.