How to Build a Real ROI Case for AI & Blockchain (Before You Invest)

Every technology vendor promises ROI. Almost none of them help you calculate it before you sign a contract. This is a problem because the question “will this project be worth it?” is the most important one a business leader can ask, and it deserves a real answer, not a marketing promise.

This guide walks you through how to build a credible ROI case for an AI or blockchain project before you commit to building it. We have included a simple framework, common cost and benefit categories, and the questions you should be asking any vendor who pitches to you.

💡 Principle:  A technology partner who cannot help you build a business case before you start probably cannot help you deliver results after you do.

Why Most ROI Projections Are Wrong

The typical vendor ROI model looks like this: take the stated problem, estimate the value of solving it, attribute 80% of that value to the proposed solution, and present a payback period of 14 months. It is almost always too optimistic, for several reasons:

• Benefits are overestimated because they assume full adoption on day one
• Costs are underestimated because integration, change management, maintenance, and retraining costs are excluded
• Risk is ignored; there is no probability weighting on the benefits actually being realised
• The counterfactual is never considered: what would happen if you did nothing, or chose a simpler solution?

A Simple ROI Framework

Step 1: Define the problem in financial terms

Before technology enters the conversation, quantify the problem you are trying to solve. Ask:

• How much does this problem cost us annually? (Include labour, errors, customer churn, missed revenue, and regulatory risk)
• How do we know this number? What data is it based on?
• What is the minimum improvement that would justify the investment?

Step 2: Map all costs, not just build costs

A common mistake is treating the project build cost as the full cost. The real cost of an AI or blockchain project includes:

• Discovery and scoping (often underestimated for complex domains)
• Data preparation and cleaning (frequently the most expensive phase)
• Integration with existing systems
• Change management and training
• Ongoing maintenance, monitoring, and retraining
• Compliance and audit costs
• Infrastructure and licensing

A useful rule of thumb: the total cost of ownership over three years is typically 2.5x to 4x the initial build cost. Any projection that ignores years two and three is misleading you.

Step 3: Model benefits conservatively

Use three scenarios:

• Conservative (50% of projected benefit realised, 6-month adoption lag)
• Base case (70% of projected benefit, 3-month adoption lag)
• Optimistic (100% of projected benefit, immediate adoption)

Weight them roughly 40% / 40% / 20% to get a realistic expected value. If your project is only profitable under the optimistic scenario, it is a high-risk investment.

Step 4: Ask the build vs. buy vs. wait question

For many AI use cases, high-quality off-the-shelf solutions now exist. Before commissioning custom development, ask:

• Is there a SaaS solution that solves 80% of this problem at 20% of the cost?
• If we wait 12 months, will an existing product improve enough to meet our needs?
• What is the competitive risk of waiting versus the financial risk of building prematurely?

A good technology partner will answer these questions honestly, even when the answer reduces the scope of their own engagement.

Step 5: For blockchain specifically, apply the blockchain necessity test

Blockchain is expensive to build, complex to maintain, and necessary in fewer situations than its proponents suggest. Before pursuing a blockchain solution, confirm that you genuinely need:

• Multiple parties who do not fully trust each other
• An immutable audit trail that cannot be managed by a central trusted party
• Decentralisation that a traditional database cannot provide

If a traditional database, a shared API, or a centralised ledger solves the problem, that is almost certainly the right answer. Blockchain has genuine, powerful use cases in supply chain provenance, cross-border payments, and digital asset management. It is not the right answer for most internal data management challenges.

Questions to Ask Any Vendor

1. Can you show me a case study from a company of my size and industry, not a large enterprise or a startup?
2. What does the total cost of ownership look like over 36 months, including maintenance?
3. What are the three most common reasons projects like this fail, and how will you mitigate them?
4. What are the measurable KPIs we will track, and what are your benchmarks?
5. What is a simpler solution to this problem, and why is it not the right choice here?

Conclusion

Building a credible ROI case before you invest is not pessimism; it is professionalism. The best technology projects are built on honest numbers, realistic timelines, and clear ownership of outcomes.