AI Project ROI: How to Calculate It Before You Invest (and Avoid the 40% That Get Abandoned)

There’s a scene that repeats itself in countless small and mid-sized companies. Someone in management comes back from a conference convinced the business needs to “do something with AI”. A project gets signed, a pilot goes live, it gets demoed in a couple of meetings and everyone agrees it looks impressive. Six months later nobody uses it, nobody knows exactly what it cost, and nobody can say whether it saved a single pound.

That isn’t bad luck or bad technology. It’s a project that started without a baseline. Working out the ROI of an AI project for an SMB before you sign anything is what separates an automation that sticks around for years from a pilot that quietly dies. And you don’t need a finance degree: you need four numbers and a spreadsheet.

Why so many AI projects get abandoned

The industry numbers are uncomfortable, but worth facing. Gartner forecasts that more than 40% of agentic AI projects will be cancelled before the end of 2027, driven mainly by escalating costs, unclear business value and inadequate risk controls (Gartner, 2025). And a study by MIT’s NANDA initiative covering 300 real deployments found that 95% of generative AI pilots produced no measurable impact on the P&L (MIT NANDA, 2025).

The revealing part of that second study is the reason why: the failure almost never sits in the model, it sits in the integration and in the choice of use case. In other words, in the decisions made before anyone writes a line of code.

The failure pattern is remarkably consistent:

• The flashiest process gets picked, not the most expensive one. A website chatbot demos beautifully; manual invoice entry demos terribly and costs £15,000 a year.
• There is no baseline. If nobody measured how long it took before, it’s impossible to prove any improvement afterwards.
• Only the licence gets counted. The real cost includes implementation, ERP integration, training, maintenance and the internal hours your own people spend on the project.
• A broken process gets automated. Automating a badly designed workflow only makes the mess happen faster. That’s why analysing the process first belongs inside the calculation, not next to it.

How to calculate AI project ROI for an SMB, step by step

The formula is the same one you already know: ROI = (annual saving − annual cost) / annual cost × 100. The hard part isn’t the formula, it’s filling it in with honest numbers. Four steps:

1. Set the baseline BEFORE you touch anything. For two weeks, measure properly: how many times a month the task happens, how many minutes each instance takes, and the loaded hourly cost of whoever does it (gross salary + employer contributions + overhead, not take-home pay). Multiply: frequency × time × hourly cost.
2. Add the cost of errors. This is the line almost everyone forgets and often the biggest one: an invoice posted to the wrong account, an order shipped at the wrong price, a deadline missed. Estimate how many errors a year and what each one costs to unwind.
3. Work out the full 12-month project cost. Implementation and integration + licences and API consumption + maintenance + your team’s internal hours. If a vendor only quotes you the licence, you’re seeing a third of the picture.
4. Calculate payback, not just a percentage. Divide the implementation cost by the net monthly saving. That number — in months — is what actually convinces a finance director. As an external benchmark, an IDC study sponsored by Microsoft puts the average return on generative AI deployments at around 13 months (IDC, 2024).

A worked example with real numbers

An accountancy practice processes 500 purchase invoices a month. Each one takes roughly 4 minutes to receive, read and key into the accounting software: 33 hours a month. At a loaded hourly cost of £22, that process costs £8,800 a year in labour alone.

With automated capture and extraction, 85% of invoices go through untouched and the remaining 15% need a 2-minute review: 2.5 hours a month, about £660 a year. Adding £1,680 a year in licences, consumption and maintenance, the process now costs £2,340 a year.

Net saving: £6,460 a year. If implementation cost £3,500, payback lands just under 7 months and first-year ROI sits around 57%, climbing above 380% from year two onwards once the setup cost is behind you.

Notice one detail: this case works because it’s 500 invoices a month. At 40 invoices a month, the exact same project never pays for itself. Volume isn’t a minor input — it’s the input.

Signs a project will NOT pay off

It’s worth stopping and rethinking when any of these show up:

• The process is infrequent. Something that happens five times a month rarely justifies an integration.
• Every case is different. If 60% of cases are exceptions requiring human judgement, AI doesn’t remove work — it shifts it to reviewing what the AI did.
• Nobody on the business side owns it. If the only sponsor is IT, usage fades as soon as the novelty does.
• Payback exceeds 18 months. Over that horizon, any change in regulation, software or org structure will wipe out the return.
• The result can’t be measured. If the expected benefit is “better image” or “staying current”, it isn’t a project — it’s marketing spend. That’s a legitimate choice, but call it what it is.

What to do with the number once you have it

The calculation isn’t paperwork to justify a purchase: it’s a prioritisation tool. Once you put five candidate processes in a table with their annual saving, their cost and their payback, the order to tackle them appears on its own. The winner is almost always a boring, repetitive process rather than the one that looked most innovative.

From there the advice stays the same: start with one case, measure it for three months and only scale what the numbers back. That is the exact opposite of buying a large platform and hunting for uses afterwards. This incremental approach is what keeps a hyperautomation strategy from stalling halfway, and it fits with keeping human oversight at the sensitive points — which also carries a cost that belongs in the sums.

Frequently asked questions

How much does an AI project cost for a small business?

A tightly scoped use case — document data extraction, an email workflow, an internal assistant — typically runs between £3,000 and £12,000 to implement, plus a monthly recurring cost for licences and consumption that usually sits between £50 and £300 for an SMB. Projects far above that range are normally covering several processes at once.

How long does it take to recover an AI investment?

For well-chosen cases, between 6 and 13 months. The IDC study cited above puts the average generative AI deployment at around 13 months. If your calculation shows more than 18 months, it’s worth checking whether you picked the right process.

How do I measure savings if nobody leaves the company?

The saving is rarely a redundancy: it’s freed capacity. You measure it in hours returned to higher-value work and, above all, in growth absorbed without hiring. “We went from 80 to 110 clients with the same headcount” is a perfectly valid — and far more honest — ROI metric.

Is it worth running a pilot first?

Yes, provided the pilot has success criteria written in advance and a fixed end date. A pilot with no metric and no deadline isn’t a test, it’s an open-ended demo — and that’s precisely the format that inflates the abandoned-project statistics.

Should we build it in-house or buy it?

The MIT study found that projects backed by specialist vendors or partners succeeded considerably more often than purely internal builds. For an SMB without its own data team, the realistic route is to lean on someone who already has the workflow running and integrate it on top of the software you already use, with no migration.

Before investing in AI, spend two weeks measuring what doing it by hand already costs you. It’s the best-returning investment in the whole project: it tells you whether it’s worth doing and, if it is, it hands you the argument to defend it.

Item	Manual	With AI
Hours spent per month	33.3 h	2.5 h
Annual labour cost	£8,800	£660
Annual licences and maintenance	£0	£1,680
Total annual cost	£8,800	£2,340
Payback on implementation (£3,500)	—	6.5 months

Contact us and we’ll analyse your case for free →