Model two groups of customers using your own assumptions. This estimates their economics; it does not identify your actual highest-cost accounts.
Nothing you type is sent to Spendline: the calculation runs in your browser.
Scenario based on your inputs · Not measured customer data.
In this scenario each of the 10 heaviest customers costs about $1,000 a month more than they pay, $10,001 in total.
Changing the inputs changes the scenario. This is not a forecast of savings from Spendline.
For comparison, ICONIQ's July 2026 survey puts the average gross margin on AI products at 53% projected for 2026, and Benchmarkit puts median SaaS gross margin at 77%.
This scenario cannot name your customers. Seeing actual cost per customer needs each AI request recorded against the customer your application says it served. Book a call to plan a guided 30-day free pilot on one workflow.
Revenue is your customer count times the average monthly price. Other cost of revenue is a flat share of each customer's revenue; the 20% starting value is an editable assumption, not a benchmark. The AI bill is split between the two groups by the share you enter, then divided evenly inside each group. Gross margin is revenue minus both costs, divided by revenue.
This calculator includes other costs so the scenario reads as gross margin. Spendline's product view is narrower: margin after AI cost is customer revenue minus attributed AI cost over the same period, where you share revenue data. Other delivery costs are not included there.
Real usage is lumpier than two groups. Within the heavy group some accounts cost more than the group average, so the single heaviest customer is at least as unprofitable as the figure shown, and more unless every heavy account costs exactly the same. The fields start with the worked example from what AI did to software gross margins; how to calculate AI gross margin by customer covers doing this properly with real per-customer cost.