A generic upgrade nag won't move someone who's already engaged and still not buying

A 0 to 100 score for the distance between how much someone uses the product and the fact they have not paid, built from sessions in the last 30 days, how broadly and deeply they use features, and email engagement, all against the profile of people who do convert. It also names the likely blocker: price, a missing feature, no authority, indecision or no urgency.

Accounts at least 30 days old, still free or trialing, scoring 70 or above. People already in a sales conversation are left out, and so is anyone who turned down an upgrade in the last 90 days.

Price gets 30% off for the first three months. A missing feature gets what most users do next plus an offer of help. No authority gets a business case template and their own usage summary to pass upward. Indecision gets a 15 minute clarity call. No urgency gets a price lock. Sent by success or by their AE.

An AI agent opens in app or by email reply, notes how long they have been using the product, and asks what is holding up the upgrade: cost, features, approval, timing or something else. Each answer routes somewhere: a discount, the product feedback queue, a business case, a later follow up, or a human.

The blocker email on day 0, an invitation to the agent on their next session if nothing happens by day 3, and on day 7 an AE task carrying the score, the blocker and a suggested approach. They leave when they subscribe, when they decline, or when the agent gets an answer and hands them on.

How the cohort grows or shrinks, conversion against a holdout left on the generic free tier nurture, conversion by blocker, how often the agent gets an answer, and how long conversion takes.