Key Takeaways
- CFOs reject AI proposals that rely on hype—build your case on specific problems, measurable baselines, and realistic projections
- Frame AI investments in CFO language: cost avoidance, efficiency gains, and risk reduction, not "transformation" or "innovation"
- Propose a limited pilot with clear success criteria before requesting full investment—it de-risks the decision and provides data for the real business case
Your CFO has seen dozens of technology investment proposals. CRM systems, productivity suites, analytics platforms, collaboration tools. Most of them got funded. Many of them delivered value.
So why does AI feel different?
It's not that CFOs are resistant to technology. It's that most AI business cases don't look like business cases. They look like trend-following. They lead with "everyone else is doing this" and "we can't afford to fall behind" rather than specific problems, measurable outcomes, and honest cost-benefit analysis.
Your CFO isn't against AI. They're against vague proposals that can't demonstrate value. Here's how to build a case that speaks their language.
Start with a Problem, Not a Technology
The worst way to start an AI business case: "AI is transforming industries and we need to adopt it."
The right way to start: "Our customer service team spends 4,200 hours per month answering the same 50 questions. Here's what that costs us, and here's a better way." Tools like AI-powered knowledge assistants can address exactly this kind of problem.
Every technology investment your CFO has approved started with a problem. AI should be no different. The problem comes first; the technology is just the solution.
Before you write anything, identify:
- What specific pain point are you addressing?
- Whose pain is it? (Which department, which roles?)
- What's the current cost of this problem? (Hours, dollars, opportunity cost)
- How do you know it's actually a problem? (What data do you have?)
If you can't answer these questions, you're not ready for a business case. You're ready for a problem definition exercise.
Establish a Credible Baseline
Your CFO will ask: "How do you know this will save money?" The answer depends on knowing what things cost now. Measuring AI ROI starts with understanding your current baseline.
Before you can project savings, you need baseline metrics. This isn't optional; it's the foundation of every credible ROI calculation.
If you don't have baseline data, get it before you proceed. Spend two weeks tracking the process you want to improve. The time invested now prevents your business case from being dismissed as speculation.
Baseline metrics to gather:
- Time spent: How many hours per week/month does this process consume?
- People involved: How many employees touch this process?
- Fully-loaded cost: What does that time actually cost including salary, benefits, and overhead?
- Volume: How many tickets, questions, documents, or transactions?
- Error rates: What percentage require rework or create downstream problems?
- Response time: How long do customers or employees wait?
Your business case gains credibility when it references real numbers from your organization, not industry benchmarks or vendor claims.
Build Conservative Projections
CFOs are professionally skeptical. They've seen plenty of projections that didn't pan out. Your projections will be scrutinized, so build them conservatively.
Rule of thumb: take your optimistic estimate and cut it in half. If that reduced number still makes the investment worthwhile, you have a strong case. If the ROI only works at optimistic projections, you have a weak case that will be questioned.
If you think AI will reduce time spent by 60%, build your business case on 30%. If the pilot shows 60%, you're a hero. If it shows 40%, you still exceeded projections. If it shows 25%, you're close to plan. You've protected yourself from three of four outcomes.
Be explicit about your assumptions:
- What adoption rate are you assuming?
- How quickly will people reach proficiency?
- What percentage of the process can realistically be automated versus augmented?
- What's the ramp-up period before you see full value?
CFOs respect honesty about uncertainty more than false precision. "We estimate 30-50% time savings based on comparable implementations" is more credible than "We'll save exactly 47%."
Speak Their Language
CFOs evaluate investments through specific lenses. Frame your proposal accordingly.
Cost reduction: Will this directly reduce operating expenses? By how much? Starting when?
Cost avoidance: Will this prevent future costs? What hiring will you not need to do? What problems will you prevent?
Productivity gain: Will this allow existing staff to do more? How will that translate to business outcomes?
Revenue enablement: Will this help close more deals, serve more customers, or accelerate time-to-market?
Risk reduction: Will this reduce compliance exposure, error rates, or operational risk?
Avoid buzzwords that mean nothing specific: "digital transformation," "AI-powered innovation," "future-proofing," or "competitive advantage." These signal that you don't have concrete value to point to.
Address the Risks Directly
Your CFO will identify risks whether you raise them or not. It's better to acknowledge them upfront and show you've thought through mitigation.
Common risks and how to address them:
Adoption risk: "What if people don't use it?"
Mitigation: Propose a pilot with users who have genuine need and willingness. Build in change management support. Set adoption milestones.
Security risk: "What about our data?"
Mitigation: Present the vendor's security credentials. Explain data handling. Involve IT and Legal early so they can vouch for the approach. Getting IT approval goes faster when you've done this homework upfront.
Cost risk: "What if costs exceed projections?"
Mitigation: Identify pricing model details. Point to budget caps or controls. Propose checkpoints where you'll reassess.
Value risk: "What if it doesn't deliver?"
Mitigation: Define clear success criteria for the pilot. Build in decision points. Make it easy to scale back if results disappoint.
Never pretend there are no risks. CFOs know better. The question is whether you've thought through the risks and have reasonable plans to manage them.
Structure the Ask
Don't ask for everything at once. Structure your request to minimize risk for the approver.
The pilot approach works well. A successful pilot builds the foundation for scaling AI from pilot to production:
- Phase 1: Limited pilot with a defined team, clear success metrics, and a defined timeline (typically 60-90 days)
- Phase 2: Expanded deployment based on pilot results
- Phase 3: Full rollout with proven ROI
For the initial ask, you're not requesting the full investment. You're requesting permission to prove the concept with contained risk. This is a much easier yes.
Frame Your Ask
- Pilot investment: $X for 90 days
- Team scope: Y employees in Z department
- Success metrics: specific, measurable targets
- Decision point: Date when we evaluate results and decide on expansion
- Exit plan: What happens if results disappoint
Show the Comparison
Every investment has alternatives. Acknowledge them.
What happens if you do nothing? What are the costs of the status quo? Are there other ways to solve this problem that are more or less expensive? Be aware of hidden costs of free AI tools when evaluating alternatives.
Your business case is stronger when it's not AI-or-nothing. Show that you've considered:
- Doing nothing (and what it costs)
- Solving the problem with people instead of technology
- Alternative technology solutions
- Partial solutions or phased approaches
When you demonstrate that you've evaluated options and AI genuinely makes the most sense, the case is more credible than if AI appears to be a predetermined conclusion.
Anticipate the Questions
Before presenting your case, role-play the CFO's questions. Common ones include:
- Why now? What's changed that makes this urgent?
- What happens if we wait six months? A year?
- Who else has done this? What were their results?
- What does IT think? Legal? Security?
- Who's going to manage this? Is this adding to someone's plate?
- What's the exit plan if it doesn't work?
- How does this compare to other investment options for the same budget?
If your CFO asked "Tell me why this isn't just hype"—what would you say? Have a clear, specific answer ready.
Include the Human Element
Numbers matter most, but people matter too. Your business case should acknowledge:
Champion: Who owns this initiative? Someone with credibility needs to stake their reputation on it.
Affected employees: How will this change their work? Have you talked to them?
IT partnership: Is IT on board? Do they have capacity?
Management alignment: Do the managers of affected teams support this?
CFOs know that technology initiatives fail for people reasons as often as technology reasons. Showing that you've addressed the organizational dynamics builds confidence.
The One-Page Summary
Your full business case might be detailed, but lead with a one-page executive summary:
- Problem: 2-3 sentences on the specific pain point
- Solution: 2-3 sentences on the proposed approach
- Cost: Investment required (especially for pilot)
- Return: Expected value (conservatively projected)
- Timeline: When you'll know if it's working
- Risk: Top risks and mitigation approaches
- Ask: What you need approved, specifically
If the one-page summary is compelling, they'll read the details. If it's not, more pages won't help.
After the Meeting
Getting approval isn't the end; it's the beginning. Plan for what happens next:
- Confirm decision criteria for the pilot in writing
- Schedule check-ins to report progress
- Commit to transparent reporting of results—good or bad
- Build the case for Phase 2 using actual pilot data
The best way to get future AI investments approved is to deliver on this one. Document results carefully. Over-deliver on communication. Make this experience a positive precedent for the next proposal.
JoySuite is designed to make the business case easy. Usage-based pricing means you only pay for what you use—no shelfware risk. Budget Shield caps ensure no surprises. And with SOC 2 certification and clear data commitments, you can check the security boxes before your CFO even asks.
