Overview

You want to invest $2M in AI across your organization. You need board approval. You need CFO buy-in. You need product and engineering alignment. You need a compelling story and solid numbers.

*Note: This lesson assumes you've completed "Calculating AI ROI" and "Total Cost of AI Ownership." The business case ties together cost calculations, ROI analysis, and stakeholder communication.*

The wrong answer is "AI is cool" or "everyone's doing it" or "we need this to stay competitive" (even if all are true). The right answer is business impact: revenue generation, cost reduction, competitive advantage. Quantified. In business language.

Here's what makes this hard: translating your technical vision into a business case that non-technical stakeholders understand and trust. CTOs and engineers think in terms of capability and architecture. Boards think in terms of returns and risk. Finance thinks in terms of NPV and IRR. Product thinks in terms of customer impact and market position. Your job is to speak all these languages and show how AI connects to each.

This lecture is about the structure, the content, and the presentation. We'll work through a real example. We'll discuss what boards actually care about (hint: it's not the technology). And we'll discuss how to handle pushback and objections.

The Business Case Structure

A strong business case for AI has a specific structure. It's not a whitepaper. It's not a technical proposal. It's a business proposal that happens to be about AI. The structure matters because it makes your thinking clear and helps non-technical readers follow the logic.

Executive Summary (1 Page, 2 Minutes to Read)

This is everything the board needs to decide. If someone reads only this page, they should understand what you're proposing, what it costs, what the return is, and what the risks are.

Template:

"We propose investing in AI capabilities across three areas: customer support automation, engineering productivity, and product personalization. Investment of $2M over 18 months ($500k Y1, $300k Y2, then $200k/year ongoing). Expected returns: break-even in 18 months, 3:1 ROI by year 3, $5M cumulative benefit over 5 years. Key risks: technology capability mismatches, implementation execution, regulatory changes. Recommend proceeding with 6-week pilot before full commitment."

That's it. Three sentences. The board either wants to drill deeper (move to next section) or make a decision. Everything else should support this summary.

The Opportunity Section (2-3 Pages)

Why are we doing this? What problem are we solving? Why now?

What problem are we solving?

"Customer support costs are growing 20% year-over-year as volume grows. We're currently at $5M/year and trending toward $6M by next year. Developers spend 40% of engineering time on boilerplate code instead of new features, limiting our shipping velocity. Competitors are adding AI-powered personalization and taking share, in our most recent customer surveys, 23% of churned customers cited lack of personalization as a primary reason for leaving."

Be specific with numbers. This section should answer: what's the current state? What's the trend? What's the cost of inaction?

Why now?

"AI capabilities have matured to the point where we can solve these problems cost-effectively. 18 months ago, this would have required building custom models. Now, APIs like Claude and GPT-4 are good enough for production use. Window of opportunity is 6-12 months before competitors catch up and this becomes table stakes."

Context matters. Why not wait? Why not do it last year? Why is now the inflection point?

Market context:

"According to recent industry surveys, a strong majority of tech leaders (92%) are using AI tools with reported productivity improvements around 1.6x. Gartner predicts companies with AI capabilities will deliver 25% of new value by 2027. 10x improvements are possible in specific high-impact domains such as customer support automation or code generation, but are not universal across all AI use cases. Competitors X, Y, Z have announced AI initiatives. Customer expectations are shifting toward AI-enhanced experiences."

Show you understand the market. Show this isn't just about us being ahead. It's about staying in the race.

The Solution Section (2-3 Pages)

What exactly are we building? Why this approach? How does it work?

What exactly are we building?

"Three integrated solutions: (1) Support automation: AI-powered ticket routing and response drafting, reducing response time from 4 hours to 15 minutes, cutting support costs by 30%. (2) Engineering productivity: AI coding assistant reducing boilerplate time from 40% to 10% of developer time, freeing up 2 FTE for new features. (3) Product personalization: ML-powered recommendations reducing churn by 2 percentage points."

Each solution should be specific: what does it do, what's the outcome, what's the improvement.

Why this approach?

"These solutions are high-impact (address our top business problems), measurable (clear before/after metrics), achievable in our tech stack (no fundamental architectural changes required), and phased (quick wins fund future investment). They're also low-regulatory-risk compared to other AI applications."

Address: why did you choose this? Why not other approaches? Why are these achievable?

Technical approach (high-level):

"Leverage Claude API for intelligence layer (reduces risk of building proprietary models, access to frontier capability, simple pricing), build custom integration layer for our workflows, invest in data infrastructure for ML personalization models. No new programming languages. No new infrastructure. Builds on existing tech stack."

Don't go deep into technical details. Board doesn't care about LLM architecture or training methodology. They care: is this achievable? Does it require fundamentally new infrastructure? Is it risky technically?

The Investment Section (1 Page)

How much does this cost? Show multi-year commitment and breakdown.

Year 1 Investment Breakdown:

• Personnel: $300k (1 ML engineer, 2 platform engineers, 0.5 PM)
  - Infrastructure/tools: $100k (compute, databases, monitoring)
  - Vendor costs: $50k (Claude API, third-party integrations)
  - Contingency (20%): $90k
  - Total Year 1: $540k

Year 2:

• Personnel: $350k (add specialist, increase PM allocation)
  - Infrastructure: $80k
  - Vendor: $100k (scaling)
  - Contingency: $70k
  - Total Year 2: $600k

Year 3+:

• Personnel: $200k/year (dedicated team of 3)
  - Infrastructure/vendor: $150k/year
  - Ongoing: $350k/year

The multi-year view shows you've thought through the full commitment. Most boards can accept investment if they understand it's not one-time. The 20% contingency shows reasonable risk planning (not too conservative, not reckless).

The Return Section (1 Page, The Most Important)

This is what the board actually cares about. What do we get back?

Support Automation Returns:

• Current cost: 500 tickets/day × $20 = $10k/day
  - Post-AI: 300 tickets/day = $6k/day (30% reduction)
  - Annual savings: $1.1M
  - Payback: less than 1 year (just from this initiative)

Engineering Productivity Returns:

• Current: 2 FTE on boilerplate (40% of 5 developers) = $500k/year
  - Post-AI: 0.5 FTE = $125k/year
  - Annual savings: $375k
  - Additional benefit: freed capacity ships X more features (estimate revenue impact)

Personalization Returns:

• Current: static experience, 3% churn rate on $50M ARR = $1.5M lost
  - Post-AI: 2% churn rate = $1M lost
  - Annual improvement: $500k (from churn reduction)
  - Upside: 8% ARPU increase from personalization = $4M revenue

Year 1 Summary:

• Total Improvement: $1.1M (support) + $375k (engineering) + $500k (churn) + $2M (conservative personalization upside) = ~$4M
  - Investment: $540k
  - Net: $3.46M positive, 6.4x return

Year 2 Summary:

• Improvements compound (support and engineering recurring)
  - Total Improvement: $1.1M + $375k + $500k + $2M = $4M
  - Investment: $600k
  - Net: $3.4M positive, 5.7x return

5-Year Cumulative:

• Investment: $540k + $600k + $350k + $350k + $350k = $2.19M
  - Cumulative Improvement: $4M × 5 years = $20M (conservative, doesn't account for competitive advantage or scaling)
  - Net: $17.8M, 8.1x return

Show the multi-year picture. Show when break-even is. Show why this is worth the investment. Most importantly: be conservative. If you show 10x return and it's only 3x, you lose credibility.

The Risks Section (1 Page)

What could go wrong? How will you mitigate it?

Technology Risk: "AI capability might not match our specific use cases or data patterns. Mitigation: 6-week pilot phase (support automation only) before full commitment. If results disappoint, we can adjust approach or stop investment with only $50k sunk cost."

Execution Risk: "Team is new to AI. Mitigation: hire experienced people, contract with consultants for implementation, monthly Go/No-Go decisions with clear criteria."

Market Risk: "Competitors move faster, capture market before us. Mitigation: we're not starting from zero, we have data advantages, 6-month timeline minimizes lag."

Regulatory Risk: "Personalization requires careful data handling. GDPR, CCPA, etc. apply. Mitigation: legal review, privacy-by-design, clear user controls."

Talent Risk: "Hard to hire ML engineers. Mitigation: identify and recruit key people now, offer competitive packages, consider consultants for initial phase."

Show you've thought about what could go wrong. Show you have mitigations. This builds credibility more than pretending there's no risk.

Pitching to the Board

Pre-Meeting Socialization

Don't surprise the board. Talk to key members before the meeting: CEO, CFO, board chair. Understand their concerns. Adjust based on feedback. If the CEO has questions, answer them before the meeting, not during. You want the board meeting to be confirmation, not education.

The Presentation (20 Minutes Maximum)

Structure:

• 2 min: Executive summary + clear recommendation ("We propose investing $2M over 18 months in AI capabilities.")
  - 3 min: Opportunity and why now (why does this matter?)
  - 3 min: Solution overview (what are we building?)
  - 5 min: Investment and timeline (how much? how long? milestones?)
  - 5 min: Expected returns and scenarios (what do we get back?)
  - 2 min: Risks and mitigations (what could go wrong?)

Do:

• Lead with business value (revenue or cost savings)
  - Be specific on ROI (show the math)
  - Show the multi-year plan
  - Acknowledge risks honestly
  - Have a clear recommendation (we propose X)
  - Be prepared for questions (anticipate them and prepare answers)

Don't:

• Get lost in technical details
  - Oversell (they'll be skeptical)
  - Use jargon without explanation
  - Present without having done homework
  - Ask for approval without giving context to understand
  - Assume non-technical board members understand AI

Dealing with Pushback:

Objection: "Why not just hire more engineers?"

Answer: "We could. It would cost more and take longer to hire and ramp. AI lets us move faster with existing team. The real answer is doing both, AI plus hiring for high-impact roles. AI handles boilerplate; new hires build new features."

Objection: "Competitors are already doing this."

Answer: "Exactly. If we wait, we'll be behind. Early movers capture customers and market share. We're at the inflection point. In 6 months, this will be table stakes. In 18 months, it will be expected. Better to invest now and lead than wait and follow."

Objection: "These ROI numbers seem high."

Answer: "You're right to be skeptical. These are based on conservative assumptions [explain]. Here are three scenarios: conservative (1.5x ROI), base case (3x), optimistic (5x). Even in conservative case, we break even in 12 months. The upside is worth the downside risk."

Objection: "How do you know you'll actually achieve these returns?"

Answer: "We don't know for certain, which is why we have a pilot phase. First 6 weeks, support automation only. We measure actual results. If they match our projections, we expand to engineering and personalization. If they're disappointing, we stop with minimal sunk cost. Go/No-Go decision at 6 weeks."

The Board Principle: Boards don't care about technology. They care about: value created, cost to create it, timeline to value, and risk. Show all four clearly. Be specific, not vague. Show you've thought it through. Have a pilot/go-no-go decision if possible.

What to Do Monday Morning

Step 1: Build the business case. Follow the structure above. Get CFO and product input on numbers. Have finance validate ROI math.

Step 2: Validate assumptions with domain experts. Support team: will 30% reduction in response time actually happen? Engineering: will engineers really spend 40% less time on boilerplate? Product: will 2% churn reduction happen? Get reality checks.

Step 3: Create the presentation deck. 10-15 slides. Executive summary, opportunity, solution, investment, returns, risks, timeline, go-no-go criteria.

Step 4: Socialize before the board. Talk to CEO, CFO, key board members individually. Get feedback. Adjust. You want no surprises at the board meeting.

Step 5: Present to board. Calm, clear, specific. Open to questions. Show you've done homework and thought through risks.

Step 6: Set up governance for execution. Once approved, establish project governance: monthly steering committee, clear milestones, go-no-go decision points (especially at 6 weeks for pilot evaluation).

FAQ: Business Case Questions

Q: What if the board asks "why not just wait and see what works, then invest?"

A: "We could. But that puts us 12-18 months behind competitors. The cost of being first is lower than the cost of being late. Plus, waiting means competitors define the market. We're better off investing now and learning fast."

Q: How detailed should the business case be?

A: As detailed as needed to make the decision. You don't need a financial model down to the dollar. You need order-of-magnitude accuracy (is it $1M or $10M return). More precision suggests false confidence.

Q: What if we don't have all the data?

A: Make reasonable assumptions and call them out. "We don't have exact support ticket numbers, but based on similar companies in our category, we estimate X." Transparency builds trust more than false precision.

Q: Should we include employee morale/satisfaction as a benefit?

A: Not in the ROI calculation (hard to quantify). Include it as a non-financial benefit: "Engineers will spend less time on mechanical work, more time on creative problem-solving. This should improve satisfaction and retention." This is valuable context but don't force it into the financial model.

Q: What if board wants us to prove it works before approving full funding?

A: Suggest a phased approach: "We'll do a 6-week pilot (support automation only, $50k cost). Measure results against our projections. If successful, approve expansion to engineering and personalization ($1.95M more). If not, we stop with minimal loss."

Key Takeaway

Business case structure: executive summary (1 page), opportunity (2-3 pages), solution (2-3 pages), investment (1 page), return (1 page), risks (1 page). Focus on ROI: investment, break-even timeline, multi-year return. Be specific, not vague. Show you've thought about risks. Present with confidence. Socialize before the meeting. Expect and prepare for objections.

The Business Case as Alignment

The best business cases aren't about getting approval. They're about alignment. You, CFO, product, engineering, and board all understanding what you're building and why. That alignment makes everything that follows easier, execution, tradeoff decisions, resource allocation.

Do the work upfront. Make the case. Build alignment. Move forward together.

On This Page

Introduction
Structure
Pitching to Board
Monday Morning Action
FAQ
Key Takeaway
Alignment & Execution

Chapter Details

Part ofCh 5: Measuring AI Impact