This lesson is being recorded

Every video of this course is being uploaded to my YouTube channel. Subscribe so you do not miss it.

Bruno AshfordBruno Ashford @brunoashfordSubscribe

Module 12 · Lesson 12.4 · 13 min

Your AI P&L

Consolidate cost, gain and return into one spreadsheet you can defend in any meeting.

Module 12
ROI: Proving It Paid Off

Turn a feeling into a number your partners will believe

Bruno Ashford
Bruno Ashford · Module 1201 / 12
The uncomfortable question
If you cannot say what it returns, someone will switch it off.
Bruno Ashford
Bruno Ashford · Module 1202 / 12
The method

Four numbers, one page

  1. 1Hours savedSame task, before and after
  2. 2True costSoftware, usage, supervision
  3. 3QualityRework and errors avoided
  4. 4AI P&LNet gain, payback, ROI
Bruno Ashford · Module 1203 / 12
Lesson 12.1

Measuring hours saved

Count human minutes, not machine minutes.

  • Time three real runs per task
  • Include your review and fixes
  • Value hours at loaded cost
  • Decide where the saved hours go
Bruno Ashford · Module 1204 / 12
Illustrative example

Roofing contractor quotes

BeforeAfter
Minutes per quote7020
Quotes per month4040
Hours per month4713

Illustrative numbers for teaching, not a sourced statistic.

Bruno Ashford · Module 1205 / 12
Lesson 12.2

What each agent really costs

The subscription is rarely the biggest line.

  • Share of your Claude plan
  • Tools bought only for the agent
  • Weekly supervision time
  • One-off build hours
Bruno Ashford · Module 1206 / 12
Lesson 12.3

Quality counts too

Fast and wrong is expensive.

  • Define failure per agent
  • Sample ten outputs a week
  • Rework above 25 percent: fix first
  • Count errors avoided as gains
Bruno Ashford · Module 1207 / 12
Lesson 12.4

The three formulas

  1. NetMonthly gainTotal gain minus total monthly cost
  2. PaybackMonthsBuild cost divided by net monthly gain
  3. ROIAnnualNet annual return divided by total annual cost
Bruno Ashford · Module 1208 / 12
Illustrative agency P&L

Five agents on one sheet

4 of 5paid back within two months
1switched off for high rework
1 pageto get partner approval
Bruno Ashford · Module 1209 / 12
Lesson 12.5

Score your next use cases

CriterionHigh score means
ValueMany hours per month
RepeatabilitySame steps every time
Data readinessInformation in one place
RiskLow damage if wrong
EffortQuick to build
Bruno Ashford · Module 1210 / 12
Decision rule

Keep, fix or kill

Every row ends with a decision.

  • Payback under three months: scale it
  • High rework: fix the context
  • Negative after three months: switch off
  • Pick three next, never ten
Bruno Ashford · Module 1211 / 12
Recap
Measure honestly, decide on one page, and let the numbers choose what you build next.
Bruno Ashford
Bruno Ashford · Module 1212 / 12
01 / 12

Use ← → to move. The marker highlights text and the pencil draws on the slide; navigation locks while you annotate. Undo removes the last mark, Esc exits.

1Open the AI ROI Calculator

Download the AI ROI Calculator from this module's materials and open it in Excel, Google Sheets or Numbers. Each row is one agent or routine. The header note explains every formula in plain English.

2Fill in gains

Enter hours saved per month and the hourly value from lesson 12.1, plus any extra gains: errors avoided, revenue from faster follow-up, a hire you did not need. Keep hard gains (cash) and soft gains (time absorbed) in separate columns.

3Fill in costs

Bring over monthly software share and supervision cost from the Cost per Agent worksheet, and the one-off build cost. Do not round up gains or round down costs. If in doubt, be conservative.

4Let the formulas do the rest

Net monthly gain equals total monthly gain minus total monthly cost. Payback in months equals build cost divided by net monthly gain. Annual ROI equals (net monthly gain times 12 minus build cost) divided by (build cost plus monthly cost times 12).

Prompt for Claude
Here is my filled AI ROI Calculator (CSV pasted below). Check every formula, flag any number that looks inconsistent or too optimistic, and write a five-line summary for my business partner: total monthly cost, net monthly gain, payback period, annual ROI, and the one agent that should be fixed or switched off.
5Decide keep, fix or kill

Any agent with negative net gain after three months of fixes should be switched off. Any agent with a payback under three months and a low rework rate deserves more volume. Write the decision next to each row so the sheet becomes a management tool, not a report.

Your accountant does not care how clever the agent is. Your partner does not want a demo. They want one page: what it costs, what it returns, and when it paid for itself.

The AI P&L is that page. Three numbers matter: net monthly gain, payback period in months and return on investment. Everything we measured in this module feeds into it.

Key points

  • One page, three numbers: net gain, payback, ROI.
  • Separate hard gains from soft gains.
  • Conservative numbers survive scrutiny, generous ones do not.
  • Every row ends with a decision: keep, fix or kill.

Avoid

  • Mixing soft time savings with cash in one number.
  • Leaving build cost out, which makes payback look instant.
  • Keeping an agent with negative return because it is impressive in demos.
  • Updating the sheet once and never again.
Real-world example: a marketing agency

A 12 person marketing agency put five agents on the sheet. Four paid back within two months, led by the monthly client report routine. The fifth, a social caption generator, looked busy but had a 40 percent rework rate and negative net gain. They killed it, moved its budget to the report routine for more clients, and the partners approved a wider rollout because the numbers were on one page.

Bruno Ashford Subscribe