The formula
ROI = (first-year revenue from won deals − full cost) ÷ full cost
A worked example (fictional)
| Line | Value |
|---|---|
| Media, 3 months at $3,000 | $9,000 |
| Management and creative, 3 months | $8,700 |
| Full cost | $17,700 |
| In-profile companies identified | 120 |
| Meetings | 14 |
| Opportunities | 6 |
| Won (33% win rate) | 2 |
| First-year value per deal | $30,000 |
| First-year revenue | $60,000 |
| ROI | 239%: revenue is 3.4 times the full cost |
Payback
Deals needed to pay back the quarter = full cost ÷ first-year deal value. Here: $17,700 ÷ $30,000 = 0.6 of a deal.
Be honest about the inputs
- Use your real win rate, not the best quarter's.
- Count only deals with an AI touch before the opportunity.
- Report a range when inputs are uncertain.
Try your own numbers in our value calculator.
A range, not a point
| Scenario | Won deals | Revenue | ROI |
|---|---|---|---|
| Low | 1 | $30,000 | 69% |
| Expected | 2 | $60,000 | 239% |
| High | 3 | $90,000 | 408% |
Fictional figures based on the example above.
Questions
Should we include lifetime value?
Report first-year value as the headline and lifetime value as a note. Boards trust the shorter figure.
What about deals still open?
Report weighted pipeline separately from closed revenue.
What if ROI is negative after one quarter?
Check the pipeline. Long cycles often turn a negative first quarter positive by the second.
Common ROI mistakes
- Counting media only. Management, creative and tools are real costs.
- Using pipeline as revenue. Report pipeline separately; ROI needs won deals.
- Ignoring time. A quarter's cost often pays back over the following quarters.
- Crediting every deal with an AI touch fully. Report influence separately from sourced revenue.
A payback view leadership understands
Instead of a percentage, show months to payback: how many months of results it takes for won revenue to cover the test's full cost. It connects directly to cash and makes the risk easy to judge.
Start here: AI ads for B2B. More articles in Measurement and reporting.
