This 3-metric board trick will stop you losing your AI budget

Let me tell you something that you might not like.

If your next board update on AI includes the phrase "the team is using it more," you are close to losing budget for your AI projects.

I say this because I see it everywhere. I've trained teams at over 100 companies, and here is what always happens.

You invest in AI, the team gets super excited, and then the board asks "so what's the ROI?" and nobody has an answer to this.

The answer they give is usually something like: "We're saving time." "People are more productive." "Adoption is up."

But this is not ROI, this is just a feeling.

And it leads to budgets getting cut.

​

​
Why "Hours Saved" Doesn't Work

Boards don't respond to time savings, because time saved that doesn't show up on the P&L is invisible.

If your team saves 5 hours a week but headcount stays the same, costs stay the same, and output quality stays the same, the board sees zero impact.

Just look at the stats.

Only 14% of 200 U.S. finance chiefs say they've seen measurable impact from AI investments. (Source: RGP 2026 CFO Research Report via CFO.com)

Only 12% of CEOs say AI delivered both cost and revenue benefits, and 56% see no significant financial benefit at all. (Source: PwC 29th Global CEO Survey, Jan 2026)

AI is probably working, but you’re measuring it in a way that people do not understand.

As KPMG's Swami Chandrasekaran (Partner / Principle AI & Digital Innovation) said:

So how do you prove your AI investments are working?

You measure it with three numbers.

​

​

The 3 Metrics That Matter

Your board already cares about forecast accuracy, reporting speed, and the bottom line. So measure AI through those exact lenses.

Forecast Accuracy: Your forecast vs actuals variance %.

Are you getting closer to what actually happens? Microsoft and Coca-Cola revamped their forecasting with AI and tightened this gap.

If your variance dropped from 12% to 7%, that's a story your board immediately understands.

Reporting Cycle Time: Days from period-end to board-ready pack.

If AI takes your close from 7 days to 4, this is 3 extra days for leadership to act on the numbers. Plus, you look like a hero in the process!

P&L Impact: The financial value finance creates for the business (not just the cost of running finance).

People always say that finance is a cost centre. But what if you could prove it's also a profit centre?

Think about this. If you use AI to identify pricing or margin issues faster, collect cash sooner, avoid late payment penalties, spot cost savings nobody else sees, or run ad-hoc analyses that directly help the business make better decisions—that's real money. That's P&L impact.

What I used to do is keep a project tracker. Every time we intervened somewhere in the business (whether it was flagging a pricing anomaly, catching a duplicate payment, or running a scenario analysis that changed a decision) we logged it. We tracked the situation before we got involved, the potential upside or cost savings, and then we assigned one of four statuses:

1. Started
2. Achieved (already in the P&L)
3. Planned for next budget
4. Failed

Just a simple log of where finance created value.

When I presented this to the team, the conversation completely changed. It wasn't "what does the finance team cost us?" anymore. It was "where else can you look?"

You only need one page with three charts. Baseline, current, trend (AI can help you with this ;))

This is your AI ROI dashboard.

You already know how to tell a financial story. You do it every time you present an ARR or a 13-week cash forecast.

So, do the same thing here for AI.

​

​

Do This Before Your Next Board Meeting

Here's exactly what to do right now.

1. Pull your baselines (30 minutes)

Go into your data and find:

• Average forecast vs actuals variance % for the last 4 quarters
• Average days from period-end to board pack over the last 3 months
• Count of finance-driven P&L interventions over the last 90 days (pricing flags, cost savings identified, cash collection improvements — anything where finance created measurable value)

This is your "before" snapshot.

Without this, you've got nothing to compare against.

2. Build your use case portfolio

Map every AI initiative across your finance domains – FP&A, close, treasury, tax, controls. Score each on impact (high/medium/low) vs effort (high/medium/low).

You want the top-right quadrant: high impact, low effort.

Be super strict. If something doesn't clearly help one of your three metrics, deprioritize it.

3. Create your business case template

Every AI initiative fills it in before getting approved:

• Problem it solves
• Options considered
• Cost (licenses, time, training)
• Expected benefit (tied to which of the 3 KPIs)
• Risks
• Recommendation

This gives the board an auditable paper trail and stops random experimentation.

4. Start monthly tracking

Measure your three KPIs every month after each AI initiative.

Don't wait for the quarterly board meeting to find out the trend is flat.

Monthly tracking gives you time to change if something isn't working, and it builds the dataset your board presentation needs.

5. Build the dashboard using HTML

Choose whichever AI tool you'd like to use and ask it to create an HTML dashboard.

It will generate an HTML file that you can download or preview (depending on the AI tool) your dashboard.

Here's how:

1. Open Gemini or ChatGPT and paste your 3 KPIs with baselines and monthly readings
2. Ask it to generate the dashboard with HTML
3. Use this prompt:

It will generate and render the dashboard in minutes.

(you can also do a similar thing using Artifacts in Claude)

Then (and this part matters) ask AI to explain how the dashboard works. You cannot present something you don't understand. Your credibility is at stake!

6. Present using "What, So What, Now What"

When you walk into the board meeting, structure it like McKinsey. Like I describe in my previous newsletter "Why nobody listens to your finance presentations (3-word fix inside)" here.

• What: Forecast accuracy improved from 12% to 7% variance. Cycle time dropped from 7 to 4 days. Finance identified 6 P&L interventions this quarter worth €340K in cost savings and revenue protection.
• So what: Our forecasts are more reliable, our reporting is faster, and finance is directly contributing to the bottom line — not just reporting on it.
• Now what: Here's our use case portfolio — these are the next 3 investments and their expected impact.

Present like this, and you've a good chance of getting the budget you want.

The One Thing to Remember

Focus on the right metrics, and present them well, and you can change the conversation.

Instead of somebody asking "why are we paying for this", they'll start to ask "where else can we use this?"

Just like with my team. I have to keep increasing my AI budget to keep up with their usage. But every week we present on how it's helping us meet our objectives, so I have no problem spending more!

As Gina Mastantuono (President & CFO, ServiceNow) said:

So give them proof.

Three numbers and three trends on one page.

Do it before your next board meeting.

​

Best,

Your AI Finance Expert,

Nicolas
​

P.S. — Are you getting budget approved for AI right now? Hit reply and tell me. I read every reply.

P.P.S. — I have covered most of the important tips for CFOs in this video 100 SECRET tips on AI for FINANCE. If you're not sure where to start, watch it. It will give you a clear picture of what finance professionals actually need to know right now.

​

​link​