The AI to EBITDA Playbook
AI does not create enterprise value until it shows up in EBITDA. The bridge is customer EBITDA: measure the dollars your AI creates on the customer's P&L, price to capture a share, and let that flow through to your own margins.
At Directly, the AI resolved support tickets at $1 per resolution. EBITDA margin went from survival-level to 22%+, sustained over two years through exit. The AI created value on the customer's income statement first. Company EBITDA followed.
Customer EBITDA Created (CEC)
CEC is the actual dollars your AI creates on the customer's P&L. Not projected savings. Not theoretical ROI. Dollars the customer CFO can measure.
An AI support agent resolves one ticket. Your customer pays its outsourced call center $5 per ticket for B2C support. Each AI resolution generates $5 CEC.
An AI sales agent closes a $150 e-commerce order. That order creates roughly $15 of customer EBITDA. That is also CEC.
Different workflows. Same measurement. If you cannot calculate CEC for your AI, you do not yet have an EBITDA story.
Customer EBITDA Return (CER)
CER is CEC divided by your price. It answers the question every customer asks at renewal: is this worth what I pay?
Below 3x: the value is too thin. You will be rationalized.
3x to 10x: durable revenue. The customer sees clear return and has incentive to renew.
Above 10x: you are underpriced. Margin is sitting on the table.
The 3x to 10x range is where AI revenue becomes structural. Price inside it.
How it works
- 01
Calculate CEC for every deployment.
- 02
Price so the customer sees 3x to 10x CER. That is the zone where revenue renews.
- 03
Build measurement infrastructure so both sides can verify outcomes. The renewal conversation becomes a P&L review, not a product review.
- 04
Let margin expand as AI delivery cost falls. At Directly, the AI agent drove roughly 50% of EBITDA at exit. That margin improvement transferred fully to the acquirer.
What it looked like at Directly
The Instant Answers AI resolved support tickets in an outcome model at $1 per resolution. Customers saw clear CER above 3x. Because pricing anchored to the customer's P&L instead of our cost structure, margin contribution stacked quarter over quarter. The AI margin improvement transferred fully to the PE-backed portfolio company.
Key takeaway
AI creates enterprise value through one path: customer EBITDA first, company EBITDA second, multiple expansion at exit. Measure CEC. Price inside the 3x to 10x CER zone. That is what an acquirer pays a premium for.
If your portfolio company has AI that isn't driving EBITDA, I've solved that problem twice.