Set the minimum quality the value depends on
An AI feature can be profitable on paper and abandoned by its users within two weeks because it gets things wrong too often. This lesson helps you set, from the outset, the quality bar its value depends on. You avoid funding a feature that nobody ends up using once the novelty wears off.
Updated November 14, 2026.
In short
The minimum quality of an AI feature is the level below which users stop trusting it and the expected value vanishes. You set it before building, along with a simple indicator you can follow from the very first tests. For a PM, this bar weighs on the decision as much as adoption or cost, because without it the value line is fiction.
Lesson objective
By the end of this lesson, you will be able to set the minimum quality without which an AI feature does not create the expected value, estimate the cost of its errors and choose the indicator, such as the acceptance rate, that will check it.
Questions this lesson answers
What quality level should an AI feature reach before launch?
The level below which the expected value no longer exists, expressed as an indicator you can track from the pilot, such as the share of suggestions users actually keep.
How much do the errors of an AI model cost?
It depends on context: time spent fixing, lost trust, even the company's liability for what its assistant says, as a 2024 court decision involving an airline chatbot showed.
Topics covered
- AI feature quality
- cost of AI errors
- acceptance rate
- LLM hallucinations
Where it fits
Estimate the value
What will the feature bring to the company, and at what minimum quality does that value exist?
Lessons in this module
- Estimate the expected value of an AI feature
- Set the minimum quality the value depends on (this lesson)
What you will learn in the course
This lesson is part of the course Build the business case for an AI feature
Related courses
- AI & ProductAdvancedRegular practice: you already use the tools or work on this topic.
Secure an AI product: prompt injection, guardrails and the AI Act
3 hr
- Tech for PMsAll levelsNo prerequisites: accessible to everyone.
Explain how a web product works, from browser to server
2 hr 30 min
- AI & ProductAdvancedRegular practice: you already use the tools or work on this topic.
Build an AI assistant for your product
3 hr