Here's a pattern I keep seeing in AI products.
A founder launches with a lovely simple plan. Usage grows. The model bill grows faster. So they switch to credits, but they pick the numbers by feel: 100 credits here, 500 there. Six months later the biggest plan is the least profitable one, and nobody can say why.
Credits only work when the rate card is built from your costs up. That's what this tool does.
How to use the Credit Pack Designer
- List the actions a customer can take and what each costs you: model tokens, GPU time, third-party APIs. If you don't know, price them with the AI Token Cost Calculator.
- Set a target margin and a credit face value. A face value of $0.01 is easy to explain: 100 credits for a dollar.
- Add your plans with a monthly price and credit allowance, or press "Suggest allowances" to give each step up a bigger bonus.
- Set a top-up premium over your entry plan's rate.
- Choose the rules: rollover, top-up expiry and what happens at zero. The tool writes the policy text for you.
How the maths works
Each action gets enough credits to hit your target margin at face value, rounded up.
Credits per action = Cost ÷ (1 − target margin) ÷ credit face value (rounded up) Price per credit on a plan = Plan price ÷ credits included Margin on an action = 1 − Cost ÷ (Credits × price per credit, ex GST)
The number that matters most is the floor: the margin when a customer on your cheapest-per-credit plan spends every credit on your most expensive action. Bulk plans give credits away cheaper, so the floor is always lower than your rate-card margin.
A worked example
The tool starts with the AI marketing studio from our credit-based pricing guide. Four actions cost from $0.006 (a social caption) to about $0.75 (a 10-step research agent). With a 75% target and $0.01 credits, the rate card comes out at 3, 15, 70 and 300 credits.
The ladder is Starter at $29 for 3,000 credits, Growth at $79 for 10,000 and Agency at $199 for 30,000. Price per credit falls from about $0.0097 to $0.0066 as customers commit.
Now the stress test. An Agency customer who spends everything on research briefs pays $1.99 per brief that costs you $0.75. That's still a 62% margin. The floor holds.
The top-up is 1,000 credits at a 30% premium over Starter, which rounds to $12.50.
How to read the result
Green is 60% and above, amber 40–60%, red below 40%. The guide suggests aiming for 70–80% on the rate card so you still clear about 60% once heavy users and bulk discounts kick in.
Watch the flags too:
- A bigger plan that costs more per credit. The guide found one real product where the $200 plan was dearer per credit than the $20 one. Upgrades should reward commitment.
- Top-ups cheaper than plans. Customers will stop subscribing and just top up.
- Harsh premiums. Clay cut its mid-cycle top-up premium from 50% to 30% in March 2026. Somewhere around 20–30% nudges heavy users to upgrade without feeling like a penalty.
And never let a meter switch a customer's work off without warning. Use alerts, a slow mode or a capped auto top-up.
What to do next
- Check the whole product's margin with light, median and heavy users in the AI Gross Margin Calculator.
- See 15 real rate cards and the rules behind them in credit-based pricing, and how one company did it in the ElevenLabs credits case study.
- Decide whether credits should sit on top of a base fee with hybrid pricing.
- Changing credits on existing customers counts as a price change. Read how to raise prices without losing customers first.
- Selling in Australia? Prepaid credits raise GST and revenue recognition questions worth raising with your accountant.
Frequently asked questions
What should one credit be worth?
Pick a face value your customers can do in their heads. One cent per credit (100 credits for $1) is popular because the dollar value is obvious, and HubSpot and Perplexity both peg credits to dollars. Abstract credits give you more room to change the rate card later, but customers find them harder to compare, so explain clearly what a credit buys.
Should unused credits roll over?
Some rollover builds trust. ElevenLabs rolls unused credits over for two months, up to three times the monthly quota, and Lovable's monthly plan credits expire two months after issue. Use-it-or-lose-it suits free daily credits. For paid plans, a short rollover with a cap is fair and stops balances piling up forever.
How big a discount should bigger plans give per credit?
Enough that upgrading feels worth it, but not so much that your floor margin falls below about 60%. In the example, Agency gives about a third more credits per dollar than Starter. Always check the floor at your cheapest credit price on your most expensive action, because that's where bulk discounts bite.
What happens when a customer runs out of credits?
You have four common options: a capped auto top-up the customer opts into, a slow mode on basic features, pay-as-you-go at the top-up rate, or a pause until the next cycle. Whichever you pick, warn customers at 80% and 100% of their allowance, and never delete their work or take their product offline without notice.
Do I charge GST on credits in Australia?
If you're registered for GST and selling to Australian customers, GST generally applies to the plan or top-up price when you sell it. Tick the GST box so the tool calculates margins on the GST-exclusive amount. Timing and recognition of prepaid credits can be tricky, so check your situation with an adviser. This is general information as at October 2026.
Built on The Subscription Playbook
This tool turns frameworks from Robert Coorey's book into something you can use in minutes. Robert is the author of The Subscription Playbook and a Director at BPAG Accountants. More about Robert.