AI ROI for small business: how to calculate it before you spend

How do you calculate the ROI of an AI project?

Estimate the yearly value (time saved, revenue gained, cost or risk avoided), subtract the full cost to build, run and supervise the agent, and compare the result against a baseline you measured before you started. Without that baseline, the ROI figure is a guess.

The maths is simple. Most small businesses skip the part that matters, the baseline, and end up with a number they cannot defend. That is a common way to fail. MIT’s Project NANDA found 95% of enterprise generative AI pilots delivered no measurable return and concluded the main cause was mismanaged adoption (Fortune). The projects that pay off do the arithmetic before they spend.

Here is the whole formula in one figure: value on one side, total cost on the other, and a baseline underneath both so the comparison means something.

What makes it honest is the baseline. ROI is always change against a starting point. If you do not know what the workflow costs today, in hours, errors or lost revenue, you have nothing to subtract from, and any percentage you quote afterwards is invented. Measure first.

How do you estimate the value of an AI project?

Estimate value in three buckets: time saved (hours removed times a loaded hourly cost), revenue gained (faster response, more capacity, higher conversion), and cost, errors or risk avoided. Use conservative numbers, count only value you will actually capture, and tie each bucket to something you can measure later.

Take the buckets one at a time. Time saved is the easiest to over-claim, so be strict: count hours the agent genuinely removes, multiply by a loaded hourly cost (salary plus overhead), and only credit time that gets reallocated to something valuable. Revenue gained comes from answering leads faster, handling more volume without hiring, or freeing your best people for higher-value work. Cost and risk avoided covers fewer errors, less rework, lower compliance exposure and contractor spend you no longer need.

Here is the method on a hypothetical example. These numbers are made up to show the arithmetic.

So the value bucket is roughly €6,000 a year before any revenue upside, and now you have a real number to set the cost against. That figure is a ceiling. You only get it if those five hours go to revenue work, and that is the step most ROI estimates skip.

What costs do small businesses forget to count?

They count the build and forget the rest: hosting, model usage, maintenance and, biggest of all, the human time to run, check and correct the agent. Add the failed-pilot tax too: the projects you abandon before one sticks. Real cost is build plus run plus supervise.

The one-time build is the number everyone fixates on, but the meter keeps running after launch. Our own prices show the split. A done-for-you agent on Qoren starts at $2,000 per workflow, fixed before we start. Running it means Qoren hosting from $39 a month (Qoren is the managed agent platform we build and run, and it bills you directly), plus model usage passed through at cost under a hard spend cap. A custom agent on your own stack gets a fixed-price quote, and maintenance is an optional flat monthly retainer quoted to your setup. Our AI agent cost guide breaks every line down, and the pricing page lists them in one place.

Supervision is the cost small businesses underestimate most. An agent that drafts replies still needs someone to read and approve them; one that enters data still needs spot checks. Budget that time honestly. It is payroll, and it stays on the books after the work moves to a model.

What is a good ROI, and when is the number lying to you?

A good ROI clears your full cost inside a payback period you would accept for any investment, often months for a focused workflow. The number lies when it counts time nobody reallocates, ignores running and supervision costs, or has no measured baseline. That is vanity ROI, a big percentage with no measurement behind it.

The cleanest way to judge a project is payback period: how many months of net value it takes to repay the total cost. A focused single-workflow build that pays back in a few months is a strong bet. A sprawling programme that “will pay off eventually” rarely does. A return you can trust has a short payback, a tight scope and a measured baseline.

Vanity ROI is the opposite. It is the impressive multiple in a vendor deck that quietly assumes every saved hour becomes billable, every cost is one-time, and the baseline was whatever makes the after number look best. The most common form is saving time that nobody redeploys, a version of Robert Solow’s old paradox: “you can see the computer age everywhere but in the productivity statistics.” We unpack that trap in the AI productivity paradox.

How do you de-risk the investment?

De-risk it in three moves: measure a real baseline before you touch any tool, pilot one narrow workflow with a clear success number, and fix the process first, because the cheapest ROI win is often a better workflow with no AI at all.

The most useful step is also the dullest: measure the baseline. Spend a week counting how long the target task really takes, how often it goes wrong and what that costs. Now your after number has something to beat, and your ROI becomes arithmetic.

Then pilot one workflow, time-boxed, with a single success metric. And before you spend on any model, ask the cheapest question in the building: could a better process fix this with no AI? McKinsey found AI high performers were roughly three times more likely to redesign their processes than other firms (McKinsey). Much of the return comes from the redesign. That is why how we work starts by mapping the workflow, and why we sometimes tell clients they don’t need AI yet. If you are buying from outside, apply the same rigour to the vendor: our guide to choosing an AI agent development company is the checklist.

If you already know the workflow, book a free 30 minute call or send a brief for a fixed-price quote to put real numbers on the cost side. If you know AI should help but not yet where, the AI consulting and audit maps your workflows, measures the baselines and ranks them by return for €1,200 to €1,800.