Experience Revolear
Explore our demos, discover our technology, get a quote, and meet our team—human and AI—in our Virtual Briefing Center.
The Problem. Sales teams judge a credit deal by its discount. But every credit a customer uses costs the seller money to deliver, so a deal can look fine on discount and still lose money.
Why It Matters. In seat-based software, a discount cuts revenue. In credit-based software, especially AI, a discount can also cut margin below what the business needs.
The Fix. Know your floor: the lowest price per credit used that still hits your target margin. Check every deal against it, including the terms that don't look like discounts.
With seat-based software, one more user costs the seller almost nothing. The code is already written. That's why traditional software has such high margins, and why sales teams could treat discounts as a revenue question only.
Credits are different. Each credit pays for real work: computing power, storage, or, in AI products, a call to a model that the seller pays for. Use more credits, and the seller's costs go up too.
That changes the math. Bessemer Venture Partners' AI Pricing Playbook (2026) puts it plainly: "Every AI query costs money (compute, inference, human-in-the-loop). You'll see 50-60% gross margins vs. 80-90% for traditional SaaS... Companies that ignore this scale to negative margins without realizing it."
So every credit deal has two questions, not one. How much revenue does it bring in? And how much does it cost to deliver?
Start with two numbers finance already has:
Put them together and you get the floor:
Floor price per credit used = Cost per credit ÷ (1 − Target gross margin)
Say each credit costs $0.40 to deliver and the target margin is 60%. The floor is $0.40 ÷ 0.4 = $1.00. Any deal where the customer ends up paying less than $1.00 for each credit they use misses the target.
Notice the words "per credit used." Costs come from usage, not from the contract. A credit the customer buys but never uses costs nothing to deliver. A credit they use for free costs the full $0.40.
The Discount Authority Matrix post followed one customer: list price $1.25 a credit, a 100K-credit commit at $1.00 (20% off), and 110K credits of uneven usage across the year. Here is the same customer, with a cost of $0.40 per credit used and a 60% target:
| Deal | Customer pays | Price per credit used | Gross margin |
|---|---|---|---|
| No commit, pays list | $137.5K | $1.25 | 68% |
| Commit at $1.00, overage $1.50, no rollover | $127K | $1.15 | 65% |
| + 30-day rollover | $115K | $1.05 | 62% |
| + lower overage ($1.10) | $111K | $1.01 | 60% |
| + 10K free promo credits | $111K | $0.93 | 57% |
Cost is $44K in the first four rows (110K credits × $0.40) and $48K in the last (120K credits). The order form says 20% off in every row with a commit. The margin falls in every row anyway. Illustrative numbers.
Three terms do most of the damage:
One term moves margin the other way: breakage. Credits that expire unused cost nothing to deliver, so a customer who commits to $100K and uses $60K looks very profitable this year. But as the Breakage post explains, that customer paid a much higher price for what they used, and will likely cut the commit at renewal. You keep the margin from breakage, but don't count on it recurring.
The floor assumes one cost per credit. Real costs move in two ways:
The math is simple once you have the inputs. The hard part is having cost per credit, expected usage and the customer's likely mix in front of the rep while they build the quote. With those built into the playbook, AI can calculate price per credit used and gross margin for each version of a deal, and flag any version that falls below the floor before it reaches the customer.
Try it this week: ask finance for your cost per credit and target margin. Calculate your floor, then check your five biggest deals: what each customer paid, divided by the credits they used. Count how many sit below it.
Once a quarter, compare margin by deal with the terms on each contract:
Most sales teams never see cost per credit. They see list price, discount and approval limits, so they negotiate the only numbers in front of them. At Revolear, we set up dozens of new order forms every quarter for usage-based businesses, and the deals that worry finance most rarely have the biggest discount. They have the most generous terms around it.
In credit-based pricing, a discount isn't just a revenue question. Every credit used has a cost, so every term that lets the customer use more for less, from rollover to overage to free credits, cuts margin. Know your floor, measure price per credit used, and check every deal against it.
Related in this series: this post is part of the Usage-Based Contracting series on credit pricing guardrails. Read more from the series:
Underpriced Minimums: The Hidden Cost of a Small Commitment
Overage Discounts: Why Cheaper Excess Undermines Renewals
Rollovers, Expiry, and Refundability: The Terms That Move Revenue
Planned Unused Credits: Breakage Is a Finance Call, Not a Sales One
Which Credit-Pricing Terms Need Approval? A Discount Authority Matrix
Raja Singh is the Founder & CEO of Revolear, the AI deal desk that runs your commercial playbook on every deal.
Sources: Bessemer Venture Partners, The AI Pricing Playbook (2026)
Explore our demos, discover our technology, get a quote, and meet our team—human and AI—in our Virtual Briefing Center.