Set AI Pilot Kill Criteria

Executive and Founder, Revenue Operations, Sales Leader, Revenue Finance.

• Before signing or starting any AI pilot.
• When a stakeholder wants to just try it and see.

• What the tool is supposed to change, in numbers
• The pilot length and cost
• Who decides go or stop at the end

Best for one task. Copy it, add your information, and run it in your AI assistant.

You are helping me design an AI pilot that can actually fail. Tool and goal: [paste]. Pilot length and cost: [paste]. Decision-maker: [paste]. Write the pilot charter: three to five measurable pass criteria with thresholds, two automatic kill conditions that end the pilot early regardless of other results, the measurement plan (what we count, from where, weekly), and the decision meeting date. Criteria must be measurable with data we actually have. If a proposed criterion cannot be measured, replace it with one that can.

Best for repeatable work. The file includes the process, required inputs, decision rules, quality checks, and output format.

---
name: set-ai-pilot-kill-criteria
description: Written pass and fail conditions for an AI pilot, agreed before the pilot starts.
license: MIT
metadata:
  author: The Revenue AI Report
  version: 1.0.0
  last-reviewed: 2026-09-04
  source: https://www.therevenueaireport.com/skills/set-ai-pilot-kill-criteria
---

# Set AI Pilot Kill Criteria

Written pass and fail conditions for an AI pilot, agreed before the pilot starts.

## When to use this skill

- Before signing or starting any AI pilot.
- When a stakeholder wants to just try it and see.

## Inputs to collect

- What the tool is supposed to change, in numbers
- The pilot length and cost
- Who decides go or stop at the end

## Process

1. State the goal as a number. If nobody can, the pilot should not start.
2. Set pass thresholds and automatic kill conditions.
3. Name the data source for every criterion. No source, no criterion.
4. Get the decision-maker to agree in writing before day one.
5. Book the decision meeting now, for the pilot's end date.

## Decision rules

- A criterion without a data source is deleted, not kept as a stretch.
- Kill conditions override everything. A tool can pass on usage and still die on complaints.
- The decision meeting date is fixed at the start and does not move.

## Output requirements

- Pilot charter: pass criteria, kill conditions, measurement plan, decision date, decider.

## Quality checks

- Every criterion has a threshold and a data source.
- At least two automatic kill conditions exist.
- The decision-maker signed it before day one.

## Limitations

- Charters fail when the decider was never really committed. Get the signature, not a nod.
- Thresholds are judgment calls. Record why you picked each number.

## Example input

AI SDR pilot, 90 days, $30k. Goal: qualified meetings. Decider: VP Sales.

## Example output

Pass: cost per qualified meeting under $400 by day 60; meeting-to-opportunity conversion within 20 percent of human SDR baseline; 80 percent of sequences need no manual repair. Kill: spam complaint rate above 0.5 percent in any week, or any send to a do-not-contact list. Measured weekly from the engagement platform and CRM. Decision meeting: day 90, VP Sales, go or stop only.

## Review checklist

- Thresholds and sources on every criterion?
- Kill conditions truly automatic?
- Decision meeting booked?

## Works with

- Playbook: Autonomous pipeline generation with a human audit lane (L5) (sales, L5) https://www.therevenueaireport.com/playbooks/autonomous-pipeline-agent-with-audit-l5
- Playbook: Make conversation data do work (L3) (sales, L3) https://www.therevenueaireport.com/playbooks/conversation-data-to-work-l3
- Playbook: Borrow the engineering harness for revenue (L3) (revops, L3) https://www.therevenueaireport.com/playbooks/engineering-harness-for-revenue-l3
- Tool: Vera (Foundation Models & Infrastructure) https://www.therevenueaireport.com/tools/vera
- Tool: Undetectable AI (Foundation Models & Infrastructure) https://www.therevenueaireport.com/tools/undetectable-ai
- Tool: Braintrust (Foundation Models & Infrastructure) https://www.therevenueaireport.com/tools/braintrust-evals

## Rules of conduct

- Write for a Director, VP, or operator. Short sentences. Explain uncommon terms.
- Separate facts from assumptions. Never hide uncertainty.
- Do not invent numbers, benchmarks, quotes, or customer names.
- Do not send messages, change CRM records, or publish anything unless the user explicitly asks.
- Flag when a decision needs human review.

## Evidence

This skill is grounded in The Revenue AI Report research:
- https://www.therevenueaireport.com/research/rollback
- Related analysis: https://www.therevenueaireport.com/blog/ai-sdr-kill-criteria-before-you-sign
- Related framework: https://www.therevenueaireport.com/frameworks/reversal-ledger



Source and updates: https://www.therevenueaireport.com/skills/set-ai-pilot-kill-criteria

1. 1.State the goal as a number. If nobody can, the pilot should not start.
2. 2.Set pass thresholds and automatic kill conditions.
3. 3.Name the data source for every criterion. No source, no criterion.
4. 4.Get the decision-maker to agree in writing before day one.
5. 5.Book the decision meeting now, for the pilot's end date.

• A criterion without a data source is deleted, not kept as a stretch.
• Kill conditions override everything. A tool can pass on usage and still die on complaints.
• The decision meeting date is fixed at the start and does not move.

• Pilot charter: pass criteria, kill conditions, measurement plan, decision date, decider.

AI SDR pilot, 90 days, $30k. Goal: qualified meetings. Decider: VP Sales.

Pass: cost per qualified meeting under $400 by day 60; meeting-to-opportunity conversion within 20 percent of human SDR baseline; 80 percent of sequences need no manual repair. Kill: spam complaint rate above 0.5 percent in any week, or any send to a do-not-contact list. Measured weekly from the engagement platform and CRM. Decision meeting: day 90, VP Sales, go or stop only.

• Thresholds and sources on every criterion?
• Kill conditions truly automatic?
• Decision meeting booked?

What does the Set AI Pilot Kill Criteria skill do?

Written pass and fail conditions for an AI pilot, agreed before the pilot starts.

Who is the Set AI Pilot Kill Criteria skill for?

Executive and Founder, Revenue Operations, Sales Leader, Revenue Finance. It sits at the start here level and takes about 30 minutes.

What do I need before I start?

Collect these first: What the tool is supposed to change, in numbers; The pilot length and cost; Who decides go or stop at the end.

What is the difference between the quick prompt and the SKILL.md file?

The quick prompt is for one task. Copy it, add your information, run it. The SKILL.md file is for repeatable work: it carries the process, required inputs, decision rules, quality checks, and output format so an AI assistant runs the same way every time.

What should I check before trusting the output?

Thresholds and sources on every criterion? Kill conditions truly automatic? Decision meeting booked?

Is it free to use?

Yes. Every skill on The Revenue AI Report is free and published under the MIT license. Attribution is welcome, not required.

• Charters fail when the decider was never really committed. Get the signature, not a nod.
• Thresholds are judgment calls. Record why you picked each number.

Run the skill, then roll it out with a playbook. Vendor links are supporting context, not a recommendation.

• Playbook: Autonomous pipeline generation with a human audit lane (L5) (sales, L5 L5 Autonomous)
• Playbook: Make conversation data do work (L3) (sales, L3 L3 Integrated)
• Playbook: Borrow the engineering harness for revenue (L3) (revops, L3 L3 Integrated)
• Tool: Vera (Foundation Models & Infrastructure)
• Tool: Undetectable AI (Foundation Models & Infrastructure)
• Tool: Braintrust (Foundation Models & Infrastructure)

• Research: rollback
• Related analysis
• Related framework