Return on AI investment (ROAI) and business case

AB-100 · Plan AI-powered business solutions · Updated

Architects are asked to prove that an AI solution is worth its price. ROAI is the structured way to do that, and the exam returns to it often.

What ROAI includes

ROAI compares total benefit with total investment over a defined period. Start with the costs, then the benefits.

  • Costs (TCO): licences, Azure consumption (models, search, storage), development and integration, data preparation, training, support and monitoring.
  • Benefits: measurable business drivers such as shorter case resolution time, fewer manual hours, higher employee productivity, lower error rates, higher conversion.
  • Strategic value: the option to reuse the platform or data for later agents.

Order matters. Quantify all development, deployment and operating costs first, then define the performance metrics, then model the benefits.

Estimating cost before you build

NeedUse
Estimate cost of planned Azure servicesAzure pricing calculator
Compare on-premises with AzureTotal Cost of Ownership (TCO) calculator
Size Fabric capacityMicrosoft Fabric SKU Estimator
See and control spend that already existsCost Management + Billing
Cut cost on steady usageAzure Reservations, savings plans

Cost Management, Reservations and savings plans act on running workloads. For a solution that does not exist yet, pick a calculator.

Build, buy or extend

Prebuilt agents (for example Microsoft 365 Copilot or Dynamics 365 agents) shorten deployment and carry low development cost. Extending with Microsoft Copilot Studio costs more but fits process-specific needs. Custom models or Microsoft Foundry agents cost the most and are justified only when the requirement cannot be met by the first two. ROAI is how you show this trade-off.

Prioritizing several initiatives

When many AI ideas compete, a simple cost and benefit table is not enough. A horizon-based framework balances quick wins (horizon 1) with scaling initiatives (horizon 2) and long-term innovation (horizon 3). A prioritization grid ranks ideas by value and feasibility, but it does not weigh long-term strategic value on its own.

Exam traps

  • Break-even analysis answers only “when do we recoup the cost”. It is not a full ROAI.
  • Cost Management + Billing, Azure Monitor and Reservations manage existing spend. They do not estimate a planned solution.
  • Brand awareness and market capitalization are not business drivers you can measure in an agent.
  • Do not start with benchmarking or CX modelling. The cost baseline comes first.
  • The cheapest option on paper is not the answer if it misses the requirement. Prebuilt first, build only when needed.

Practice questions

Choose an answer, then review the explanation to check your understanding.

1. A company wants to judge whether an AI service desk assistant is worth building. The analysis must include total cost of ownership, expected savings from automation and the option to extend existing AI. What do you include?

Show answer

Answer: B.ROAI weighs the full cost (TCO) against measurable benefits such as savings, and can compare build, buy or extend options. Break-even alone ignores strategic value.

2. You are starting an ROAI analysis and it must be accurate. What do you do first?

Show answer

Answer: A.An accurate ROAI starts from a complete cost baseline. Without it, the benefits cannot be compared to a real investment figure.

3. Which pair of business drivers is measurable in an ROAI analysis of a support assistant?

Show answer

Answer: B.Use drivers you can measure inside the solution, such as handle time and productivity. Brand awareness and market cap are too indirect.

4. You must estimate the monthly Azure cost of a sentiment analysis solution before it is built. Which tool do you use?

Show answer

Answer: C.The Azure pricing calculator (or the TCO calculator for migration comparisons) estimates cost for planned services. Cost Management shows spend that already exists.

AB-100 course outline

All 6 lessons · Course overview
  1. Return on AI investment (ROAI) and business casePlan AI-powered business solutions Free
  2. Data readiness and grounding for agentsPlan AI-powered business solutions Free
  3. Choosing the right agent type and platformDesign agentic AI solutions Full access
  4. Copilot Studio design, escalation and Foundry integrationDesign agentic AI solutions Full access
  5. Monitoring and measuring agent qualityDeploy and manage AI business solutions Full access
  6. Agent ALM, deployment and governanceDeploy and manage AI business solutions Full access