AI Feature Attribution

How to use it

1. Enter your total ARR, the fraction of users who use the AI features, the ARPU uplift those users show, your monthly AI infra cost, the churn difference for AI users, and your total user count. The tool separates your user base into AI and non-AI cohorts and back-solves the base ARPU, so it isolates how much of your revenue the AI features actually earn rather than letting AI take credit for revenue it did not move.
2. Read the ARR attributable to AI (the ARPU premium times the AI users), the net figure after annual AI infra cost, the AI cohort gross margin, and the ARR kept by lower churn; the line under the tiles shows the AI and non-AI user counts, both ARPUs and the annual infra cost. The net AI-attributable ARR is the honest number, because it credits AI only for the incremental revenue and then subtracts what the AI infrastructure costs to deliver it.
3. Use the net figure to answer whether the AI features pay for themselves. It is common for an AI feature to show an impressive ARPU uplift that mostly evaporates once the infra cost is netted out, and this tool surfaces that gap directly instead of letting a gross uplift number justify the spend on its own.
4. Read the ARR kept by lower churn as a second source of value, reported separately so it is not double-counted. If AI users churn meaningfully less, the retained revenue can rival the direct ARPU uplift, which reframes the AI feature from a revenue add-on into a retention investment. Test a conservative uplift and a higher infra cost together to see how robust the attribution is before you defend the feature's budget.
5. Re-run as AI adoption spreads through your base, as infra costs change, or as you measure real cohort churn rather than estimate it. Pair this with the agent cost per validated customer tool for the acquisition-side economics, so you see both what AI features cost to run and what they earn across the full customer lifecycle.

Related calculators

• Run the numbers Agent Cost Per Validated Customer

  AI and infra spend per activated retained user, with gross margin at any subscription price.

• 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.

Articles

• 9 min read AI Feature Attribution: Pulling ARR Lift Out of the Noise

  Split ARR uplift from an AI feature using cohort ARPU deltas, infra cost, and churn difference. A worked $1.2M ARR example with honest sensitivity bands.

• 13 min read The Method for Measuring AI Feature ROI

  Method for measuring AI feature ROI has three layers: marginal cost, revenue attribution, retention impact. Usage rate is the wrong primary metric.

• 14 min read Build vs Buy in 2026: When the LLM API Kills Your Moat

  Build vs buy in 2026: which AI product features survive when Anthropic and OpenAI ship the same thing, and why distribution moats outlast a model wrapper.

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

By situation

• Plan your AI stack
• Price your product
• Know your runway

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/ai-feature-attribution.js";

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