---
title: "What is the most common mistake in calculating AI ROI?"
description: "Ignoring the human review and correction time that remains after launch, followed closely by not taking a baseline before the build."
canonical: https://sudosapient.si/answers/what-is-the-most-common-mistake-in-calculating-ai-roi
updated: 2026-10-27
publisher: Sudo Sapient (https://sudosapient.com/)
---
# What is the most common mistake in calculating AI ROI?

**Ignoring the human review and correction time that remains after launch, followed closely by not taking a baseline before the build.**

The usual mistake is counting gross time saved per task and multiplying by volume, while ignoring the time people spend reviewing outputs, fixing errors and handling the cases the system routes back. A second mistake is skipping the baseline, which makes any improvement claim unverifiable. Measure net hours and full running cost against a baseline taken before the build.

## Go deeper

- [How to Measure AI ROI Without Fooling Yourself](https://sudosapient.si/entries/measuring-ai-roi)
- Terms: [AI ROI](https://sudosapient.si/terms/ai-roi)

## More on getting to production

- [Why do AI proofs of concept fail to reach production?](https://sudosapient.si/answers/why-do-ai-proofs-of-concept-fail-to-reach-production)
- [What does 'in production' mean for an AI system?](https://sudosapient.si/answers/what-does-in-production-mean-for-an-ai-system)
- [How do you measure AI ROI?](https://sudosapient.si/answers/how-do-you-measure-ai-roi)