Solutions

Understand AI customer profitability

AI companies need customer revenue and actual delivery costs in the same view. AI customer profitability is that view: what each customer pays, what it costs to serve them, and the margin that remains.

Connect what you earn with what it costs to deliver AI — at the customer, task, and workflow level.

Why revenue and delivery cost belong together

A growing book of business can still leak margin. Two customers on the same plan can generate very different AI delivery costs, so revenue alone does not tell you which accounts are healthy.

AI customer profitability analysis asks a narrower question than a cloud bill: for this customer, in this period, did what we earned cover what it cost to deliver their AI work?

AI gross margin per customer

AI gross margin per customer is gross profit divided by that customer’s revenue, where gross profit is revenue minus delivery cost for the same period.

How to calculate AI gross margin: subtract attributed delivery cost from customer revenue, then divide by revenue. A customer with $1,000 of revenue and $600 of delivery cost has $400 of gross profit and a 40% gross margin. A second customer with the same revenue and $920 of delivery cost has an 8% gross margin. Same price, different economics.

The figures above are a synthetic illustration, not a customer result.

AI customer contribution margin

AI customer contribution margin is customer revenue minus the variable delivery costs you attribute to that customer — typically model calls, tool calls, retries, and evaluations.

Teams differ on whether shared infrastructure belongs in the figure. Pick one explicit rule and keep it stable. The distinction that matters is between cost tracking (what was spent) and profitability analysis (whether the revenue attached to a customer covered that spend).

How to connect AI usage costs with customer revenue

Use the same customer identifier, period, and currency on both sides. Usage costs that cannot be tied to a customer stay in an unattributed bucket instead of being silently dropped or spread without a rule.

Then read margin at three grains: the customer, the workflow, and the task. How to measure AI workflow profitability is the middle grain — a customer can look acceptable overall while one workflow is underwater. The delivery-cost side of that comparison is AI cost-to-serve.

What Arbiter Margin is for

Arbiter Margin connects customer revenue with AI delivery costs so founders, finance leaders, and product teams can see customer and workflow margins.

It works beside billing, observability, and accounting. It does not replace those systems, and it does not claim a measured savings rate.

Continue with how AI cost-to-serve rolls model, tool, and workflow costs into one figure.

FAQ

How is this different from an ARR dashboard?
ARR shows commercial activity. AI customer profitability adds delivery cost so gross margin per customer is visible.
Is contribution margin the same as gross margin?
They are related. Contribution margin usually subtracts variable delivery costs. Gross margin follows whatever your finance policy treats as the cost of delivering the AI work. State the rule so the two figures are not mixed.
Who is this for?
AI startup founders, CTOs, CFOs, and product managers at AI agent and AI SaaS companies who need revenue and delivery cost in one view.

From the blog