Customer MarginsPer End User
Calculate cost-to-serve at the request level. Report margin by customer, feature, and model. Updated continuously.
Per-customer margin data
Why AI products benefit from customer-level cost attribution
1
Provider invoices show total spend
OpenAI, Anthropic, and AWS bills show aggregate costs. They don't show cost per customer.
2
Revenue systems show total revenue
Stripe shows payments and subscriptions. It doesn't connect revenue to the costs incurred to serve each customer.
3
Margin calculation requires joining these data sources
Per-customer margin = customer revenue - attributed costs. This requires request-level cost tracking.
4
Bear Billing automates this data pipeline
SDK captures request metadata. Dashboard calculates and displays margin by customer, feature, and model.
What this enables
Pricing decisions based on actual cost-to-serve data rather than aggregate estimates.
What Margin Analytics calculates
Customer margin by end user, feature, and model. Updated continuously from request-level data.
Continuous
Updated as requests are processed
Per Customer
Margin calculated for each customer
Drill-Down
By feature, model, and time period
1
Customer Margin Table
Revenue, cost-to-serve, and margin for each customer. Sort and filter by any dimension.
Customer A$200 revenue | $12,000 cost | -$11,800 margin
Customer B$500 revenue | $413 cost | +$87 margin
Customer C$100 revenue | $112 cost | -$12 margin
2
Margin Over Time
Historical margin data by customer, feature, or model. See trends and compare periods.
58%
October
42%
November
-16 pts
Change
3
Margin Breakdown by Dimension
See margin contribution by feature, model, customer tier, or custom metadata labels.
Margin by Model
GPT-4o: 57.8% margin | Claude Sonnet: 61.8% margin | Custom Model: 69.1% margin
Common Questions
Request-level cost attribution for AI products
Customer margin by end user, feature, and model. Updated continuously.