Write the go/no-go criteria and design the pilot
Once a project is underway, almost every team finds a good reason to keep going, even when results disappoint. This lesson teaches you to set the rules of the game while there is still time. Your pilot becomes a real test, and its conclusion is accepted by everyone before a single result comes in.
Updated November 14, 2026.
In short
A reliable go/no-go decision on an AI project rests on quantified thresholds and a date, agreed at the outset, when nobody has a reason to soften them. The pilot then checks the most uncertain assumptions. For a PM, this is the best protection against a project that carries on simply because so much has already been spent on it.
Lesson objective
By the end of this lesson, you will be able to write quantified go/no-go criteria with a deadline for an AI feature before building it, and design the pilot that tests the most fragile assumptions.
Questions this lesson answers
How do you make a go or no-go decision on an AI project?
Choose indicators tied to the fragile assumptions, give each a threshold and a date, then agree with the decision-maker on what happens in each case.
How do you design the pilot for an AI feature?
Limit its scope and duration, and build it to measure the assumptions profitability depends on, rather than to put on a demo for stakeholders.
Topics covered
- AI go/no-go decision framework
- kill criteria
- AI pilot
- AI project decision
Where it fits
Profitability and decision
From what adoption level does the feature pay for itself, and how do you decide without fooling yourself?
Lessons in this module
- Compute the margin and break-even adoption
- Build three scenarios and find the assumptions that flip the decision
- Write the go/no-go criteria and design the pilot (this lesson)
- Present the business case to your leadership
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 & ProductExpertAdvanced level: for deeper mastery and specialization.
Ship and monitor an AI feature in production
3 hr
- AI & ProductAdvancedRegular practice: you already use the tools or work on this topic.
Get reliable, product-ready AI outputs (JSON, schemas, function calling)
2 hr 30 min
- AI & ProductExpertAdvanced level: for deeper mastery and specialization.
Design reliable agents and tool calling
3 hr 30 min