The Real ROI of AI for Small Business (and How to Calculate It)
By Alex Carlson
"AI will save you time" is true and useless. It doesn't tell you whether a $200/month tool returns $400 or $40. To decide where to spend, you need to put a real number on the return — and most small businesses either skip the math entirely or trust a vendor's rosy claim.
This guide gives you the actual ROI formula, shows you the three places AI returns come from, accounts for the costs everyone forgets, and sets realistic expectations for payback.
The formula
AI ROI uses the same formula as any other investment:
ROI = (Benefits − Costs) ÷ Costs × 100
Simple. The trick is being honest about both sides — especially costs, which are usually bigger than the subscription price, and benefits, which are easy to inflate.
Get the costs right (all of them)
The sticker price is only part of it. A realistic first-year cost looks like:
Total first-year cost = (monthly subscription × 12) + (setup hours × hourly value) + (training hours × hourly value) + lost productivity during rollout
For a lot of small businesses, the time lines exceed the software line. A $30/month tool is $360 a year — but if it takes 20 hours to set up and train the team on, at a $50/hour blended value, that's another $1,000. Skip these and your ROI math is fiction.
The three places AI returns actually come from
For small businesses, AI ROI shows up in three buckets. Estimate each in dollars:
1. Time saved on manual tasks
The most common and easiest to quantify. Hours per week saved × number of people × hourly value × 52. If an AI tool saves your front desk 5 hours a week at a $25/hour value, that's $6,500 a year — against a tool that might cost a few hundred.
2. Revenue from captured leads
Often the biggest bucket and the most overlooked. Leads that previously went to voicemail or got a slow follow-up and went cold. If AI lead-response captures even a handful of deals a year that you'd otherwise lose, that revenue usually dwarfs the tool cost. This is where AI frequently pays for itself many times over.
3. Avoided hiring costs
If AI lets you handle growth without adding a role — or handle the work of a role you'd otherwise need to backfill — the avoided salary is real ROI. For a small business, deferring one hire can be tens of thousands of dollars.
What payback actually looks like
Done well — one focused use case, costs counted honestly — many small businesses see full payback within 30 to 60 days. That speed is the whole argument for piloting one use case at a time: you get a real ROI number fast, then use it to justify the next purchase. Broad, all-at-once rollouts bury the signal and delay the payback.
Don't trust a single number — model the range
Here's the nuance most ROI calculators miss: your return isn't one number, it's a range. Lead capture might be great or mediocre; adoption might be fast or slow. A serious estimate models a pessimistic, expected, and optimistic scenario (often called P10 / P50 / P90) so you can see the downside, not just the dream case. If the pessimistic scenario still clears your cost, it's a safe bet. If only the optimistic one does, it's a gamble.
That's exactly what the Rémis ROI simulator does — it runs the scenarios on server-computed financials instead of vendor hype, so the numbers are auditable.
A worked example
A 12-person services business considers an AI tool at $300/month:
- Costs: $3,600 subscription + ~$1,500 setup/training = $5,100 year one.
- Benefits: 6 hrs/week saved across 2 staff at $30/hr = $18,720; plus 3 extra deals captured at $1,500 = $4,500. Total ≈ $23,220.
- ROI = (23,220 − 5,100) ÷ 5,100 × 100 ≈ 355%. Payback in well under two months.
Change the assumptions and the answer changes — which is the point. Model your own numbers before you buy.
Where Rémis fits
Rémis turns this from a back-of-napkin guess into a defensible projection: the ROI simulator models your scenarios on real cost data, and the full strategy report ties the financials to a specific implementation plan. Both run on locked, server-computed figures the AI can't fabricate. See the plans or start with a free readiness audit to make sure the use case is viable first.
Frequently asked questions
How do you calculate AI ROI for a small business? Use ROI = (Benefits − Costs) ÷ Costs × 100. Count all costs — subscription, setup, training, and lost productivity during rollout — and estimate benefits across three buckets: time saved, revenue from captured leads, and avoided hiring.
What's a good ROI for AI tools? Many small businesses see payback within 30–60 days on a well-chosen first use case, which implies a strong annual ROI. The key is that even a pessimistic scenario should clear your costs; if only the optimistic case does, treat it as a gamble.
What costs do people forget when calculating AI ROI? Setup time, training time, lost productivity during the switch, and usage-based pricing that climbs as you scale. These often exceed the subscription cost itself.
Where does AI ROI come from for small businesses? Three places: time saved on manual tasks, revenue from leads that no longer slip away, and hiring you can avoid. For most small businesses, captured-lead revenue is the largest bucket.
Written by Alex Carlson, founder of Rémis (University of Miami, BBA Finance + BBA Business Technology). Methodology reflects published 2026 AI ROI guidance for small businesses.