ROI of AI: How to Calculate It and What to Realistically Expect

Every business leader considering AI asks the same question: "What's the return?" It is the right question. But most of the answers online are vague — "AI can save you money" or "AI boosts productivity." That is not useful. You need a framework you can apply to your own numbers.

This guide gives you the formula, the benchmarks, and a real case study you can use to model your own AI ROI. If you are trying to build a business case for AI, this is the page to bookmark.

How to Calculate AI ROI: A Step-by-Step Formula

AI ROI is not mysterious. It follows the same logic as any capital investment. You add up the value it creates, subtract the cost, and divide by the cost. Here is how to do it precisely.

Step 1: Identify the Target Workflow

Start with one workflow. Not "AI for the whole company" — one process that is manual, repetitive, and high-volume. Common starting points include data entry, report generation, lead qualification, invoice processing, and client communication. If you are not sure where to start, a readiness assessment helps identify the highest-value target.

Step 2: Measure the Current Cost

Quantify what the workflow costs today. You need three numbers:

• Time cost: Hours per week spent on the task, multiplied by the fully loaded hourly rate of the people doing it
• Direct costs: Contractor fees, software subscriptions, or outsourced services tied to the workflow
• Error cost: Revenue lost or rework required due to manual mistakes

Example: If three people each spend 5 hours per week on a task at a $75/hour loaded rate, that is $58,500 per year in time cost alone.

Step 3: Estimate the AI-Driven Improvement

Be conservative. Assume 60-80% automation of the workflow, not 100%. Account for the time humans still need to review, approve, or handle edge cases. Also estimate any new revenue the automation enables — for example, faster lead response times that increase close rates.

Step 4: Calculate Total Investment

Your total AI investment includes:

• Consulting and development: The cost of designing and building the solution (typically $10,000-$75,000 depending on complexity)
• Implementation: Integration with existing systems, data preparation, testing
• Training: Time for your team to learn the new workflow
• Ongoing costs: API fees, maintenance, updates (usually 10-20% of the build cost annually)

Step 5: Apply the Formula

A result of 2.0 means you get $2 back for every $1 invested. A result of 3.5 means $3.50 back per $1. Anything above 1.0 is a positive return.

AI ROI Benchmarks by Industry

Your expected ROI depends on your industry, the complexity of the workflow, and how well the AI solution is matched to the problem. Here is what the data shows.

McKinsey's 2025 Global AI Survey found that companies in the top quartile of AI adoption reported profit margin improvements of 5-10 percentage points. The bottom quartile saw little to no measurable return — typically because they invested in AI without a clear target workflow or success metric.

What Separates High-ROI From Low-ROI AI Projects

The data consistently shows three differentiators:

1. Specificity: Projects targeting a single workflow outperform broad "AI transformation" initiatives by 3x (McKinsey, 2025)
2. Data readiness: Companies with clean, structured data in the target area see ROI 2x faster than those who need extensive data preparation
3. Executive commitment: Projects with a named owner who has authority to change workflows are 4x more likely to reach full deployment (Deloitte, 2025)

The Dig Solutions Case Study: $50K Investment, $198K Annual Savings

This is Third Coast AI's own case study. We built custom AI agents for Dig Solutions, a digital marketing agency managing paid media campaigns across multiple platforms and clients.

The Problem

Dig Solutions was spending 200+ hours per year on manual reporting, data aggregation, and campaign analysis across Google Ads, Meta Ads, and other platforms. The work was done by a mix of full-time staff and contractors. It was accurate but slow, and it prevented the team from spending time on higher-value strategic work.

The Investment

• Consulting and build: $40,000 for custom AI agent development
• Integration: $7,000 for connecting to existing ad platforms and data sources
• Training and rollout: $3,000
• Total: $50,000

The Results

The $198,000 in annual savings breaks down as follows: $78,000 in recaptured staff time (200 hours at blended rate), $84,000 in reduced contractor spend (60% reduction from $140,000 annual contractor budget), and $36,000 in revenue from new capacity — the freed-up time allowed the team to take on additional clients without hiring.

At $198,000 in annual value against a $50,000 investment, the ROI is 3.96x in year one and the payback period is approximately 17 months when you account for the ramp-up period.

Common ROI Mistakes (and How to Avoid Them)

Mistake 1: Only Counting Time Saved

Time saved is the most obvious benefit, but it is often the smallest. Cost reductions (fewer contractors, fewer tools, less rework) and revenue gains (faster turnaround, new capacity, better quality) can be 2-3x the value of time savings alone. Count all three.

Mistake 2: Ignoring Ongoing Costs

AI is not a one-time purchase. API costs, model updates, maintenance, and occasional retraining are real. Budget 10-20% of the initial build cost per year for ongoing operations. If you skip this, your ROI projections will be inflated.

Mistake 3: Measuring Too Early

Most AI projects need 2-3 months of tuning before they hit steady-state performance. Measuring ROI at week two is like judging a new hire on their first day. Give the system time to be optimized, and measure at the 6-month and 12-month marks.

Mistake 4: Comparing AI to Perfection Instead of the Status Quo

The question is not "Is the AI perfect?" The question is "Is the AI better than what we are doing now?" If your current process has a 5% error rate and the AI has a 2% error rate, that is a meaningful improvement — even though it is not zero.

When AI Does Not Deliver ROI

AI is not always the answer. Here are the scenarios where AI typically fails to deliver a positive return:

• The problem is not well defined. If you cannot clearly describe the inputs, outputs, and decision logic of a workflow, AI cannot automate it. Define the process manually first.
• Data quality is poor. AI learns from your data. If your data is incomplete, inconsistent, or siloed across systems that cannot be connected, the AI will underperform. Data cleanup must come first.
• The organization will not change. AI changes workflows. If the team is unwilling to adopt new processes, even a technically excellent AI solution will sit unused.
• The scope is too broad. "Automate everything" projects fail. "Automate invoice processing" projects succeed. Narrow the scope to one workflow at a time.

A good AI consulting partner will tell you when AI is not the right solution. If someone is selling you AI for every problem, they are selling, not consulting.

How to Measure AI ROI on an Ongoing Basis

Calculating ROI at launch is step one. Tracking it over time is what separates companies that scale AI from those that abandon it. Here is a practical measurement framework.

Monthly Metrics

• Hours automated: Track the actual hours the AI handles vs. what humans previously spent
• Error rate: Compare AI output accuracy to the previous manual baseline
• Cost per task: Divide total AI operating costs by the number of tasks completed

Quarterly Metrics

• Cumulative savings: Running total of time, cost, and revenue value generated
• Payback progress: How close you are to recouping the initial investment
• Scope expansion opportunities: New workflows that could benefit from the same AI infrastructure

Annual Review

• Total ROI: Full-year value divided by total investment (including ongoing costs)
• Comparison to projection: How actual results compare to the original business case
• Next investment decision: Based on results, where to invest next

For a deeper look at measurement frameworks, see our guide on how to measure AI ROI over time.

Frequently Asked Questions

Next Steps

If you are building a business case for AI, start with the formula above and one target workflow. If you want help identifying the right workflow and modeling the ROI for your specific situation, take our AI readiness assessment or learn about our consulting process.