The AI Revenue Leak Checklist

Find where AI activity is failing to become revenue.

Twelve questions across the six operating decisions.

Twelve questions. Ten minutes. Scored Controlled, Leaking, or Exposed. No email required to see your result.

The same twelve checks I run before any engagement. If your score is clean, you will know. If it is not, you will know exactly where to look.

Start the checklist 12 questions · ~10 min · no email to see your score
The scoring

The scoring, up front

Every question has three answers worth 0, 1, or 2 points. Zero means the step is controlled. Two means it is leaking. The total lands you in one of three bands:

  • 0–8 Controlled. Your system is mostly decided. Re-run this quarterly.
  • 9–16 Leaking. Real money is moving through unmade decisions.
  • 17–24 Exposed. The system is running you, not the other way around.

No trick questions, no inflated scoring to scare you into a call. The bands are the same ones I use in paid work.

Prefer paper? Download the one-page PDF and score it by hand.

The checklist

Twelve questions

Answer from what you can see today. If you are not sure, choose the middle answer. Uncertainty is part of the diagnosis.

Where the tools live

What should AI do? · What should go into the CRM?

01 Is there a written line for what AI and automation are allowed to do on their own, versus only prepare work for a person to finish?
02 Can you say what each tool in the revenue stack is responsible for?
03 Do tool outputs, the drafts, enrichment, and call notes, enter the CRM as structured fields?

Where the workflow hands off

What should humans review? · What should not be automated?

04 When a tool drafts something customer-facing, does a named person approve it before it goes out?
05 When automation hits a case it was not built for, an unusual reply or a lead that fits no rule, does it route to a person?
06 Have you decided on purpose which steps stay human?

Where the revenue should show up

What should trigger follow-up? · What should go into the CRM?

07 Can your team see, in one place, which leads are waiting on follow-up right now?
08 When a lead goes quiet, does a defined trigger bring it back to a person?
09 Is there one agreed record where a deal's real status lives?

Where leadership reads the score

What should leadership measure? · What should not be automated?

10 If leadership asked today for the return on the tool stack, how long would it take to produce the number?
11 Do you know whether your tools added review or rework hours that did not exist six months ago?
12 If a critical tool went down today, would the revenue motion still run?

Your score appears here once all twelve are answered. It stays in your browser.

The fit call

Before you book

What it is

Fifteen minutes, three questions. They're about what happens if nothing changes, where that costs you, and what return you could show leadership today. You leave knowing whether a Decision Sprint fits and what I'd map first.

What it is not

Not a pitch deck. Not a discovery call that turns into a sequence. Not for AI-curious startups without revenue, and I will say so on the call rather than waste your fifteen minutes.

The Crow's Signal
Book a fit call

Elsewhere

LinkedIn YouTube (when live) Newsletter (when live)