Chapter 14 — Building the Business Case¶

Part III — AB-731 track: Leading AI Transformation

In 30 seconds¶

The core idea: funding AI means matching a solution to a need, understanding the cost drivers (tokens, licenses, adoption effort), and choosing the right licensing model.
Why it matters: business value and licensing thread through AB-731 (Domains 1 and 3).
The exam angle: expect questions on selecting a solution, tokens/ROI, and Copilot vs Foundry licensing models.
Remember: Microsoft 365 Copilot = per-user subscription; Foundry = consumption (pay-as-you-go or commitment tiers).

Exam map¶

Exam map — AB-731 · Domain 1: business value of generative AI · Domain 3: licensing and cost

1. Key concepts¶

A transformation leader must justify AI spend. That means connecting a solution to a measurable business outcome, then choosing the commercial model that fits.

📌 Key concept: value comes from scale and automation (Chapter 1) — many people doing repetitive knowledge work faster and more consistently. The business case weighs that value against total cost.

Cost drivers¶

Licenses — the flat, predictable cost of Microsoft 365 Copilot (per user, per month).
Consumption (tokens) — the variable cost of Foundry models: input + output tokens × price (Chapter 1).
Adoption & governance — change management, training, and oversight (Chapters 15–16) — real costs that determine whether value is realized.

\[\text{ROI} = \frac{V - C}{C}\]

where V is the value created (time saved, quality gains, new capability) and C is the total cost.

2. How it works — licensing models¶

The exam expects you to distinguish the two commercial worlds.

	Microsoft 365 Copilot	Foundry Tools
Model	Per-user subscription	Consumption
Options	Monthly / annual; included with certain Microsoft 365 plans; Copilot Chat pay-as-you-go for some agent usage	Pay-as-you-go or commitment tiers
Predictability	Flat, predictable per head	Variable with usage; commitment tiers discount steady workloads
Best for	Broad employee productivity	Custom apps, spiky or specialized workloads

🎯 Exam tip: per-user subscription → Microsoft 365 Copilot; pay-as-you-go / commitment tiers → Foundry. A predictable per-employee rollout is a subscription; a variable custom workload is consumption.

🔍 How it works: choose subscription when usage is broad and steady (every knowledge worker); choose consumption when usage is variable or the solution is bespoke. Commitment tiers trade flexibility for a lower rate on predictable Foundry usage.

Selecting a solution to meet a need¶

🎯 Exam tip: "select a generative AI solution" rewards the simplest fit — adopt Microsoft 365 Copilot for productivity; extend for gaps; build on Foundry only for bespoke needs (Chapter 12). Don't over-engineer.

3. In the real world¶

Scenario — two very different bills. A company rolls out Microsoft 365 Copilot to 500 knowledge workers: a predictable per-user monthly cost it can budget precisely, justified by hours saved on drafting and summarizing. Separately, it builds a customer-facing app on Foundry whose cost rises and falls with customer traffic — pay-as-you-go, later moved to a commitment tier once volume stabilizes to cut the rate. Same company, two commercial models, each matched to the workload.

4. Exam tips¶

🎯 Exam tip: remember the mapping cold — Copilot = per-user subscription (monthly / included); Foundry = pay-as-you-go or commitment tiers.

🎯 Exam tip: ROI isn't just cost — include the value (time saved, quality, new capability) and the adoption cost needed to realize it.

🎯 Exam tip: tokens (input + output) drive consumption cost, not the Copilot per-user price.

5. Common pitfalls¶

⚠️ Pitfall: assuming per-user pricing applies to Foundry, or token pricing to Microsoft 365 Copilot. They're different commercial models.

Ignoring adoption cost: a license nobody uses has negative ROI; budget for change management.
Over-building for cost's sake: bespoke Foundry solutions cost more to build and run than adopting Copilot.
Cost-only thinking: a cheaper option that doesn't deliver value is not the better business case.

6. Practice questions¶

1. An organization wants to give all 1,000 employees Copilot in their Microsoft 365 apps with a predictable cost. Which licensing model applies?

A. Pay-as-you-go tokens
B. Per-user (monthly) Microsoft 365 Copilot subscription
C. Foundry commitment tier
D. Free web chat only

Answer

Correct: B. Microsoft 365 Copilot is licensed per user per month — predictable for a broad rollout. A and C are Foundry consumption models; D wouldn't ground in work data at scale.

2. A custom Foundry app has highly variable usage. Which pricing best fits initially?

A. Per-user subscription
B. Pay-as-you-go (consumption)
C. A one-time license
D. It cannot be priced

Answer

Correct: B. Variable custom workloads suit pay-as-you-go; a commitment tier can lower the rate once usage is steady. Per-user subscription is the Copilot model; C and D are incorrect.

3. Which best reflects a sound AI business case?

A. Choose the cheapest option regardless of outcome
B. Weigh value created (time saved, quality, new capability) against total cost, including adoption
C. Ignore adoption and governance costs
D. Always build custom

Answer

Correct: B. ROI balances value against total cost, including the adoption effort to realize it. A and C ignore value/cost realities; D over-builds.

4. What primarily drives the cost of a consumption-based Foundry model?

A. The number of employees
B. Input and output tokens processed
C. The number of slides created
D. The office location

Answer

Correct: B. Consumption pricing is driven by tokens (input + output). Headcount drives Copilot subscription cost, not Foundry consumption; C and D are irrelevant.

Chapter 14 — Building the Business Case¶

In 30 seconds¶

Exam map¶

1. Key concepts¶

Cost drivers¶

2. How it works — licensing models¶

Selecting a solution to meet a need¶

3. In the real world¶

4. Exam tips¶

5. Common pitfalls¶

6. Practice questions¶

Further reading¶