Use when a credit model needs designing: action-to-credit translation, tier structure, and the enterprise rate-card play. Produces a credit schedule, tier recommendations, and negotiation guidance.
Translate actions into credits
- t-shirt size by complexity and value delivered, never by compute cost. Small / medium / large / XL. Reference points: a quick document review ~3 credits; a full multi-step assessment ~200.
- pick the denomination for psychology, not math: 100,000 credits feels abundant; 100 feels stingy. Scale the schedule so allowances read as generous.
- keep the schedule simple enough that a buyer develops intuition — if every action needs a lookup, collapse sizes.
Design the tiers
- 80/20 rule: each tier includes a credit allowance sized so most customers in it stay within bounds. Credits exist to make expansion frictionless, not to nickel-and-dime.
- quote new customers at 70–80% of expected usage — growing into more credits is a win conversation; unused mountains of credits are a churn signal.
- design for ~80% utilization with a ~20% buffer. Allow limited rollover; unlimited rollover erases the expansion signal.
- no penalty-priced overages. Every credit mechanic should be designed to make the customer feel like they're winning.
Structure the enterprise rate card
- keep the rate card out of the contract, referenced from it — a living document that evolves as the product improves, so shipping new capabilities never reopens procurement.
- lock the rate card values themselves (what each action costs in credits) and keep them consistent across accounts. When an enterprise buyer wants a lower effective rate, adjust credit volume, not the rate card — that's a bundling conversation, a sales job.
- buyers aren't looking for transparency; they're looking for a vendor they can trust. Consistency builds that; bespoke per-account rate cards destroy it.
Sell it in human terms
Never lead with the abstract unit. Lead with outcomes, then translate:
- "your plan covers ~500 contract reviews a month — what a junior paralegal handles in a week"
- human value equivalents: "100,000 credits ≈ one person full-time for a month"
- pitch against future hiring needs, not existing staff — a growth story, not a threat
- back it with visibility: real-time consumption dashboard, alerts before thresholds, and a monthly line like "80,000 credits consumed → 2,400 documents processed, $18,000 cost avoidance"
What good looks like
A great credit package lets the vendor ship new workflows for years without a single contract renegotiation, while every consumed credit maps to a value unit the buyer can see. Mediocre packaging prices credits off compute, buries the schedule in the contract, and punishes overage — recreating bill-shock, the thing credits exist to kill. Watch for: allowances set so low that customers hit the wall in month one, denominations that make normal usage feel expensive, and discounting the rate card instead of bundling volume.
MUST size credits on value delivered, not cost incurred. NEVER embed the full rate card in the contract. NEVER design a mechanic where the customer winning feels like the vendor losing.
1---2name: credit-packaging3description: Use this skill when the user is moving an AI product to credits or structuring a credit-based enterprise deal — "how many credits should this action cost," "design our credit tiers," "should the rate card go in the contract," "how do we explain credits to buyers." Translates agent actions into a credit schedule, designs tiers and allowances, and structures rate cards for enterprise negotiations. Built on Manny Medina's credit-selling playbook at Paid: credits are the bridge between subscriptions and outcomes — sell the outcome, meter the credit, and design every mechanic so the customer feels like they're winning.4---56Use when a credit model needs designing: action-to-credit translation, tier structure, and the enterprise rate-card play. Produces a credit schedule, tier recommendations, and negotiation guidance.78## Translate actions into credits910- t-shirt size by complexity and value delivered, never by compute cost. Small / medium / large / XL. Reference points: a quick document review ~3 credits; a full multi-step assessment ~200.11- pick the denomination for psychology, not math: 100,000 credits feels abundant; 100 feels stingy. Scale the schedule so allowances read as generous.12- keep the schedule simple enough that a buyer develops intuition — if every action needs a lookup, collapse sizes.1314## Design the tiers1516- **80/20 rule**: each tier includes a credit allowance sized so most customers in it stay within bounds. Credits exist to make expansion frictionless, not to nickel-and-dime.17- quote new customers at 70–80% of expected usage — growing into more credits is a win conversation; unused mountains of credits are a churn signal.18- design for ~80% utilization with a ~20% buffer. Allow limited rollover; unlimited rollover erases the expansion signal.19- no penalty-priced overages. Every credit mechanic should be designed to make the customer feel like they're winning.2021## Structure the enterprise rate card2223- keep the rate card **out of the contract, referenced from it** — a living document that evolves as the product improves, so shipping new capabilities never reopens procurement.24- **lock** the rate card values themselves (what each action costs in credits) and keep them consistent across accounts. When an enterprise buyer wants a lower effective rate, adjust credit volume, not the rate card — that's a bundling conversation, a sales job.25- buyers aren't looking for transparency; they're looking for a vendor they can trust. Consistency builds that; bespoke per-account rate cards destroy it.2627## Sell it in human terms2829Never lead with the abstract unit. Lead with outcomes, then translate:3031- "your plan covers ~500 contract reviews a month — what a junior paralegal handles in a week"32- human value equivalents: "100,000 credits ≈ one person full-time for a month"33- pitch against future hiring needs, not existing staff — a growth story, not a threat34- back it with visibility: real-time consumption dashboard, alerts before thresholds, and a monthly line like "80,000 credits consumed → 2,400 documents processed, $18,000 cost avoidance"3536## What good looks like3738A great credit package lets the vendor ship new workflows for years without a single contract renegotiation, while every consumed credit maps to a value unit the buyer can see. Mediocre packaging prices credits off compute, buries the schedule in the contract, and punishes overage — recreating bill-shock, the thing credits exist to kill. Watch for: allowances set so low that customers hit the wall in month one, denominations that make normal usage feel expensive, and discounting the rate card instead of bundling volume.3940MUST size credits on value delivered, not cost incurred. NEVER embed the full rate card in the contract. NEVER design a mechanic where the customer winning feels like the vendor losing.
swan-gtm