Spot upsell candidates from usage data
When to use: Run it at the start of a quarter, or any week the team needs a pipeline of expansion conversations rather than guesses.
Prompt
You are a role for company, a product description. Our plans work like this: plan structure. Review the account data below and identify the count accounts most likely to benefit from an upgrade right now. Look specifically for signal, and ignore accounts that show no pressure against their current limits. Here is the data: usage data. Return the answer as output format. For each account, state the single strongest piece of evidence from the data, name the tier they should move to, and write one sentence an account manager could open a conversation with. Do not invent numbers that are not in the data, and flag any account where the evidence is thin.
How to adapt it
- Replace usage data with a real export -- the more columns you paste, the sharper the ranking.
- Change signal to whatever actually predicts expansion in your product, not what you wish predicted it.
- Adjust count down to five if you want a list a rep will actually work this week.
Why it works
It gives the AI a role, the plan logic it needs to reason about tiers, and real evidence to point at. Constraining the output format and forbidding invented numbers turns a vague brainstorm into a list an account manager can pick up and use without checking every row by hand.