KNOWLEDGE BASE · GUIDE
The 12-Month AI Business Case, in the CFO's Format
A 12-month AI business case in the CFO's format has five sections: hours times loaded cost, three scenarios, payback period, worst case stated plainly, and a one-page summary. The exact section order matters because finance teams already read in this sequence. Build the case before you ask, and defend it line by line (skillpilotadvisory.ai/scan, verified 15 Oct 2026).
The case is not a vendor report. It is your numbers, in the format your CFO expects, with the assumptions stated and the worst case not hidden. Most AI programmes fail the board test because the case was never built in this format (skillpilotadvisory.ai/about, live 15 Oct 2026).
The five sections, in order
| Section | What it contains | What a CFO looks for |
|---|---|---|
| 1. Hours and cost | Workload hours per week, loaded cost per hour, total annual cost of the work today | A named source for the hours. "Estimated" is a flag. Timesheet or service log is not. |
| 2. AI share | The percentage of the workload AI could absorb, stated conservatively | A number below 50%. A 30% reduction that holds for twelve months beats a 70% claim that collapses in quarter two. |
| 3. Three scenarios | Conservative, base, and aggressive. Each runs the 12-month model with different AI shares and adoption rates | The conservative scenario as the decision basis. If the conservative case does not pay back, the base case does not matter. |
| 4. Payback period | The month the cumulative savings exceed the cumulative cost | A single month number. Not a range. The CFO wants to know when the money comes back. |
| 5. Worst case | What happens if the AI share is zero. What happens if the sponsor leaves. What happens if the vendor raises prices. | Stated plainly, not buried. The worst case is the section that earns the CFO's trust. |
How Scan builds it
Scan takes your workload hours and costs and builds the 12-month model with three scenarios, a payback period, and a CFO summary. It delivers as a spreadsheet pack you can defend line by line (skillpilotadvisory.ai/scan, verified 15 Oct 2026).
The readiness map pre-fills the case. Your answers about data, people, approval and sponsor shape the scenarios. If your readiness score is low on people, the adoption rate in the conservative scenario drops. The case reflects your reality, not a vendor's template.
The monthly proof engine closes the loop. Two months of CSV become the one line that moved, judged against your own definition of normal. The case predicted a number. The proof reports whether the number showed up.
What this does not cover
The business case does not guarantee the outcome. It models your own numbers and states the assumptions. If the hours estimate is wrong, the case is wrong. If the loaded cost is wrong, the payback period is wrong. The honesty is in the inputs, not in the model. AI never sets a score, changes a number, or invents a cause (skillpilotadvisory.ai/scan, FAQ, live 15 Oct 2026).
Sources and context
- Scan's business case: workload hours and costs become a 12-month model with three scenarios, payback period, and CFO summary. US$149 or RM640, one-time (skillpilotadvisory.ai/scan, verified 15 Oct 2026).
- Most AI programmes fail the board test: spend rises while cost, attrition, revenue and cycle time barely move (skillpilotadvisory.ai/about, live 15 Oct 2026).
- Malaysia's National Guidelines on AI Governance and Ethics (MOSTI, published Nov 2024) name accountability as an expectation. The worst-case section is where accountability starts.
- Malaysian enterprises often sequence AI top-down before defining the problem, which turns adoption "into an unnecessary expense" (Manminder Kaur Dhillon, CEO, Supernewsroom.ai, World AI Show Malaysia 2026, The Rakyat Post, 14 Sep 2026). The business case is the bottom-up answer.
FAQ
How long does it take to build the case? The model runs in minutes once you have the hours and costs. The time is in finding three honest numbers: weekly hours, loaded cost, and the AI share you will defend.
What AI share should I use? Start at 30% for the conservative scenario. If the case does not pay back at 30%, do not raise the share to make it work. Fix the readiness factor that is blocking adoption.
Does the CFO need the full model? No. The CFO needs the one-page summary: hours, cost, AI share, payback period, worst case. The full model is for the person defending the line.
What if the payback period is over 12 months? State it. A payback period over 12 months is not a rejection. It is a signal to fix the readiness factor that is driving the delay, then re-run the case.