The new book from Dan Leiva

DELIVER

The Operator's Playbook for Governing AI in Customer Experience

AI now makes decisions your customers live with. DELIVER shows you how to put a named owner behind every one of them.

  • Publishing November 10, 2026
  • From the author of AMPLIFIED, a #1 Amazon bestseller
Use the free calculator Order the book
DELIVER book cover by Dan Leiva
Cost per Bad Decision Calculator

Find out what your AI's wrong decisions really cost.

Most AI business cases count the savings and leave out the decisions the AI gets wrong. Those costs land in other budgets: repeat contacts, recovery work, refunds, and customers who leave. This calculator puts them back on the same page, using the method from Chapter 9 of DELIVER.

Enter your numbers for one decision type. You get your cost per bad decision, your break-even error rate, and a one-page return you can hand to your CFO.

The fields start with the illustrative example from Chapter 9. Replace them with your own numbers. Your numbers stay in your browser and are never sent anywhere.
How the calculator works

The arithmetic from Chapter 9, on your numbers.

Cost per bad decision = repeat contacts × cost per contact + recovery time × loaded hourly cost + share remediated × remediation amount + share who leave × lifetime value.
Net value per month = decisions × savings per decision − decisions × error rate × cost per bad decision.
Break-even error rate = savings per decision ÷ cost per bad decision. Above this rate, the bad decisions cost more than the automation saves.
Class 2 comparison. In DELIVER, a Class 2 decision is one where the AI recommends and a person decides. The calculator reruns your numbers with the share of savings you would keep and the lower error rate you would expect.

The results are estimates based on the numbers you enter. They are not financial advice. Treat them as estimated until a pilot or finance review proves them.

About the book

Every metric was green. The customer was still gone.

AI agents now answer customers, issue refunds, and make decisions that used to belong to people. Most companies can tell you how many contacts their AI handled. Few can tell you who owns the decision when it gets one wrong.

DELIVER is the operator's playbook for closing that gap. Dan Leiva draws on 25 years running customer operations at Apple, Intuit, eBay, and Travelers to lay out a seven-stage method for putting AI in front of customers with named owners, clear boundaries, and a gate that nothing passes without evidence.

Inside, you'll learn how to:

  • Decide which decisions AI can make alone, which need a person, and which it should never touch
  • Name an owner for every automated decision before it reaches a customer
  • Measure the cost of a bad decision alongside the cost per resolution
  • Protect the frontline people who handle what the AI can't
  • Prove the financial return to a CFO who wasn't in the room
  1. D
    Diagnose
    Map the decisions your AI makes and where they fail customers.
  2. E
    Engineer
    Name the experience, draw the decision boundary, then design the screen.
  3. L
    Lock
    The gate. Nothing goes forward without a named owner.
  4. I
    Incubate
    Pilot in real conditions, with kill criteria set in advance.
  5. V
    Validate
    Prove the model holds at scale, across every system a decision touches.
  6. E
    Embed
    Keep ownership working through drift, model updates, and turnover.
  7. R
    Return
    Prove value a CFO trusts without hollowing out the experience.
About the author

Dan Leiva

Dan Leiva has spent 25 years running customer operations and the technology behind them at Apple, Intuit, eBay, and Travelers. At eBay, his teams raised AI resolution from 67 percent to more than 91 percent while customer satisfaction more than doubled. He is the founder of CXAmplify, the author of AMPLIFIED, a #1 Amazon bestseller, and a United States Marine Corps veteran.

More about Dan

DELIVER publishes November 10, 2026.

AI doesn't remove the need for an owner. It removes the places an unowned decision can hide.

Order the book Book Dan to speak about DELIVER