Agent Cost Per Validated Customer

How to use it

1. Enter your monthly active users, your activation rate, your 30-day retention rate, the tokens each user consumes per month, your model's input and output prices per million tokens, your infra cost per user, and optionally your price per user. The tool splits token usage evenly between input and output, so enter total monthly tokens; with current pricing such as Claude Sonnet 5.5 at $2 input and $10 output per million, or Claude Haiku 5.5 at $0.10 and $0.50, the model you choose dominates the cost line.
2. Read the cost per validated customer, total monthly cost (split into AI tokens and infra under the tiles), and the count of validated customers (active users times activation times retention). A validated customer is one who activated and stuck around for 30 days, so this metric divides your full variable cost by the users who actually matter, not the raw signups that token-burn alone would flatter.
3. Use cost per validated customer rather than cost per signup, because the gap between them is where AI products quietly lose money. If only a third of users activate and half of those retain, your real acquisition cost can be six times the naive per-user figure, and a price that looked profitable against signups can be underwater against the customers who stay.
4. Enter your price per customer and read the gross margin on paying customers: revenue from validated customers only, against AI and infra spend for every active user. When the margin is thin or negative, the highest-leverage fixes are usually a cheaper model tier for routine calls, prompt and token reduction, or improving activation and retention so the same AI spend is amortised across more validated customers.
5. Re-run whenever model pricing changes, when your activation or retention shifts, or when you consider a price change. Pair this with the model price drop stress test to see how falling token costs reshape these economics, and the AI product margin tool for the full per-user cost picture.

Related calculators

• Run the numbers Runway With AI Cost Shock

  Stress-test runway against an LLM vendor price hike with break-even revenue trajectory.

• Run the numbers Free Trial to Paid Target

  Reverse-solve trial→paid conversion rate from revenue target, signups, and ARPU.

• Run the numbers AI vs Human Support Cost

  Compare AI-first and human-only support cost with token spend and escalation overhead.

• Run the numbers AI Feature Attribution

  ARR attributable to AI features, net of infra cost, with cohort gross margin and retention lift.

Articles

• 12 min read Prompt Caching ROI: When It Pays Off, When It Does

  Anthropic prompt caching mechanics, hit-rate sensitivity, and the break-even derivation. Worked example: 61-84% input cost reduction on a docs assistant.

• 13 min read Pricing AI Features: Cost-Plus vs Value-Based

  Pricing AI features: cost-plus breaks because token cost per customer can vary 100-fold; a per-seat base plus a usage cap holds margin and room for value.

• 11 min read AI Support Tool Pricing 2026: 7 Vendors Ranked by Cost Model

  AI support tool pricing 2026: eesel (from $299/mo for 500 tickets), Gorgias ($0.90/resolution), Intercom Fin ($0.99); Decagon, Sierra, Ada are quote-only.

More in AI product economics

Understand the costs, margins, and pricing of building AI-powered products.

• Startup Cost Estimator
• Profit Margin Calculator
• Profit Margin / Markup / Discount Calculator
• SaaS Pricing Strategy Calculator

Use it from code

The same calculation as a JavaScript module you can import. It runs where you import it, with no request, key or rate limit.

import { compute } from "https://aibizhub.io/engines/agent-cost-per-validated-customer.js";

Input and output contract, TypeScript types and the guide for agents.