An AI agency had a strong champion at a healthcare network — the VP of Patient Operations was convinced that AI-powered scheduling would save $400,000 annually. She had seen the demo, loved the approach, and was ready to sign. Then she hit the wall: a $180,000 annual engagement needed CFO approval, and the CFO wanted a formal business case. The champion tried to write one herself and submitted a three-paragraph email with rough estimates. The CFO rejected it within a day: "I need real analysis, not back-of-napkin math." The AI agency stepped in, built a 12-page business case with detailed cost modeling, sensitivity analysis, implementation risk assessment, and a phased investment plan. The CFO approved the budget two weeks later. The agency founder later said: "We lost eight weeks because we did not build the business case upfront. Now we build it before the champion even asks."
Most AI deals are not lost because the prospect does not want to buy. They are lost because the prospect cannot get the budget approved internally. Your champion is convinced, but the CFO, the CEO, or the procurement committee needs evidence that justifies the investment. If your champion has to build that evidence themselves, you are relying on someone who does not have your expertise, your data, or your time to make your case. Build the business case for them, and you remove the single biggest obstacle between interest and signed contract.
Why Business Cases Matter More for AI
AI Investment Carries Perceived Risk
AI is still viewed as experimental by many finance leaders. Unlike established technology categories (ERP, CRM, cloud infrastructure), AI does not have decades of proven ROI data. This means AI budget requests face higher scrutiny than other technology investments. Your business case must be more rigorous, more specific, and more conservative than business cases for established technologies.
The Numbers Must Survive Finance Scrutiny
CFOs and finance teams are trained to find holes in financial arguments. Optimistic projections, missing assumptions, and vague estimates trigger immediate skepticism. Your business case must be built with the same rigor that finance professionals apply to their own analyses.
Multiple Stakeholders Need Different Evidence
The business case must satisfy different audiences simultaneously:
- CFO/Finance: Needs NPV, payback period, and sensitivity analysis
- CEO: Needs strategic rationale and competitive positioning
- CTO/IT: Needs technical feasibility and integration assessment
- Operations/Business Owner: Needs operational impact and implementation plan
- Procurement: Needs vendor comparison and contract justification
A single document that addresses all these perspectives dramatically increases approval probability.
Your Champion Is Not a Financial Analyst
Your internal champion may be brilliant at their function — operations, marketing, customer service — but they are typically not trained in building financial business cases. When they try to create one, the result often lacks the financial sophistication that CFOs expect. By building the business case yourself, you ensure it meets the standard required for approval.
The Business Case Framework
Section 1: Executive Summary (1 Page)
The executive summary must communicate the entire business case in one page. Many approval decisions are made based on the executive summary alone — the detailed sections are reviewed only if the summary generates interest.
Include:
- The problem statement: One paragraph describing the specific business problem, quantified with current costs and impact.
- The proposed solution: One paragraph describing the AI solution at a high level.
- The financial summary: Key financial metrics in a simple table:
- Total investment over 3 years
- Total projected savings/revenue over 3 years
- Net present value (NPV)
- Payback period
- Return on investment (ROI)
- The recommendation: A clear statement recommending the investment with a requested approval amount and timeline.
Section 2: Current State Analysis (2-3 Pages)
Document the prospect's current situation with specificity and evidence:
Process description: How the current process works, who is involved, what tools are used, and what the workflow looks like step by step.
Cost analysis: The fully loaded cost of the current process:
- Direct labor costs (hours x fully loaded hourly rate)
- Error and rework costs (error rate x cost per error)
- Technology and tool costs (current software, maintenance)
- Opportunity costs (what the team could be doing instead)
- Risk costs (compliance exposure, customer churn, competitive disadvantage)
Performance metrics: Current performance on relevant KPIs:
- Processing time per unit
- Error rate
- Customer satisfaction scores
- Capacity utilization
- Compliance incident rate
Use the prospect's actual data wherever possible. Data from discovery conversations, shared reports, and public sources. Where actual data is unavailable, use industry benchmarks clearly labeled as estimates.
Section 3: Proposed Solution (2-3 Pages)
Describe the AI solution in business terms, not technical terms:
What it does: Describe capabilities in terms of business outcomes, not technology features. "Automatically processes and categorizes 85% of incoming invoices" not "uses NLP and computer vision to extract data from unstructured documents."
How it integrates: Describe how the AI solution works with their existing systems. Reference specific platforms they use.
Implementation approach: Phased implementation plan with clear milestones:
- Phase 1 (Weeks 1-4): Setup, integration, and initial training
- Phase 2 (Weeks 5-8): Pilot deployment with measured subset
- Phase 3 (Weeks 9-12): Full deployment and optimization
- Ongoing: Monitoring, optimization, and expansion
Resource requirements: What the client needs to invest beyond your fee:
- Staff time for training and adoption (estimate hours)
- IT support for integration (estimate hours)
- Change management effort (estimate hours)
- Any infrastructure or technology requirements
Section 4: Financial Analysis (3-4 Pages)
This is the section that CFOs scrutinize most carefully.
Cost of the proposed solution:
| Cost Category | Year 1 | Year 2 | Year 3 | Total | |--------------|--------|--------|--------|-------| | AI agency fees | $144,000 | $144,000 | $144,000 | $432,000 | | Internal implementation costs | $35,000 | $5,000 | $5,000 | $45,000 | | Infrastructure costs | $12,000 | $12,000 | $12,000 | $36,000 | | Training and change management | $15,000 | $5,000 | $5,000 | $25,000 | | Total cost | $206,000 | $166,000 | $166,000 | $538,000 |
Projected benefits:
| Benefit Category | Year 1 | Year 2 | Year 3 | Total | |-----------------|--------|--------|--------|-------| | Labor cost reduction | $180,000 | $240,000 | $240,000 | $660,000 | | Error cost elimination | $85,000 | $95,000 | $95,000 | $275,000 | | Revenue from freed capacity | $50,000 | $120,000 | $150,000 | $320,000 | | Risk reduction value | $30,000 | $40,000 | $40,000 | $110,000 | | Total benefits | $345,000 | $495,000 | $525,000 | $1,365,000 |
Net benefit analysis:
| Metric | Value | |--------|-------| | Total 3-year investment | $538,000 | | Total 3-year benefit | $1,365,000 | | Net benefit | $827,000 | | ROI | 154% | | Payback period | 9 months | | NPV (at 10% discount rate) | $612,000 |
Assumptions list: Document every assumption in the financial model. CFOs respect transparency about what is assumed versus what is proven:
- Staff fully loaded cost: $45/hour (based on client-provided data)
- Error rate reduction: 75% (based on comparable client results, conservative estimate)
- Capacity freed by AI: 1,200 hours/year (based on current process analysis)
- Revenue per freed capacity hour: $85 (based on client's average billable rate)
- Year 1 benefits begin at month 3 (allowing for implementation ramp-up)
Section 5: Sensitivity Analysis (1 Page)
Show what happens if the projections are wrong. This is what separates a professional business case from an optimistic guess.
Scenario analysis:
| Scenario | Benefit Reduction | 3-Year Net Benefit | ROI | Payback | |----------|-------------------|---------------------|-----|---------| | Best case | 0% | $827,000 | 154% | 9 months | | Base case | -20% | $554,000 | 103% | 12 months | | Conservative | -40% | $281,000 | 52% | 18 months | | Worst case | -60% | $8,000 | 1.5% | 34 months |
"Even in the conservative scenario, where benefits are 40% below projections, the investment achieves a positive NPV and a 52% return. Only in the worst case scenario — where benefits are 60% below projections — does the investment approach break-even. Based on our experience with comparable implementations, the base case scenario is the most likely outcome."
Break-even analysis: "The investment breaks even if the AI system delivers at least 39% of projected benefits. Based on our track record and the specifics of this implementation, we have high confidence in exceeding this threshold."
Section 6: Risk Assessment and Mitigation (1-2 Pages)
Address risks proactively:
Implementation risks:
- Risk: Integration with existing systems takes longer than planned. Mitigation: Phased implementation with milestone reviews.
- Risk: Staff adoption is slower than expected. Mitigation: Dedicated change management plan with training and support.
- Risk: AI accuracy does not meet targets initially. Mitigation: 90-day optimization period included in the engagement.
Ongoing risks:
- Risk: AI performance degrades over time. Mitigation: Continuous monitoring and quarterly model retraining.
- Risk: Vendor dependency. Mitigation: All systems built on standard technology with full documentation for portability.
- Risk: Regulatory changes affecting AI use. Mitigation: Ongoing regulatory monitoring included in the engagement.
Section 7: Alternative Analysis (1 Page)
Compare the proposed AI solution against alternatives the approval committee might suggest:
Alternative 1: Do nothing. Cost: $X per year in continued inefficiency, plus competitive risk.
Alternative 2: Hire additional staff. Cost: $X per year for additional headcount, with limitations on scalability and consistency.
Alternative 3: Implement basic automation (non-AI). Cost: $X for automation tools, with limitations on handling complex cases and adaptability.
Alternative 4: Build AI capability in-house. Cost: $X for hiring AI engineers, data scientists, and infrastructure, with 12-18 month timeline to first deployment.
Show that the proposed AI agency engagement is the most cost-effective path to the desired outcome.
Presenting the Business Case
Preparing Your Champion
Your champion will likely present the business case internally. Prepare them thoroughly:
- Walk them through every section and help them anticipate questions
- Prepare a FAQ document with answers to the 10 most likely questions from finance
- Create a one-page summary they can use for hallway conversations
- Offer to attend the presentation or be available by phone for technical questions
Handling Finance Objections
"These projections are too optimistic." "The sensitivity analysis on page [X] shows the investment is positive even at 40% below projections. We used conservative assumptions throughout and have documented every assumption for your review. We are happy to adjust any specific assumption you believe should be more conservative."
"Can we start smaller to test the ROI?" "Yes. The phased implementation plan includes a pilot phase that tests the AI on a subset of the workload. We will measure actual results against projections at the 90-day mark. If results fall short, we can adjust or exit before committing to the full deployment."
"We need to compare this against other vendors." "We encourage that. The business case is structured around the business outcomes, not our specific technology. You can use the same framework to evaluate any vendor's proposal. We believe the comparison will favor our approach based on [specific differentiators]."
"What happens if we cancel the engagement?" "The contract includes a termination clause with [X] days notice. Your internal systems and data remain fully under your control. We provide complete documentation so that you can maintain or transition the AI systems independently."
Building Business Cases at Scale
Create Templates by Vertical
Develop business case templates for each industry and use case you commonly sell. Templates should include:
- Standard financial model structure with industry-specific assumptions
- Typical cost categories and benefit categories for the vertical
- Industry benchmarks for key metrics
- Common risk factors and mitigations
Templates reduce business case preparation from 20-30 hours to 8-12 hours while maintaining quality.
Collect Client Data for Benchmarks
As you build business cases for more clients, collect anonymized data on actual results:
- Actual versus projected ROI
- Actual implementation timelines
- Actual adoption rates
- Actual performance improvements
This real-world data strengthens future business cases because you can reference proven results, not just projections.
Your Next Step
Take your most recent proposal that is pending approval and build a formal business case using the framework above. Include the financial analysis with NPV and payback period, the sensitivity analysis, and the risk assessment. Send it to your champion with a note: "I built this to help with your internal approval process. It is designed to address the questions your finance team will ask. Would it be helpful to walk through it together before you present it internally?" The business case will accelerate the deal, and the process of building it will become a reusable capability for every future engagement.