What does AI cost for an SME? An honest explanation of the price ​

What does AI cost for an SME? The honest answer is: it depends on the task. But that is no excuse to stay vague. This page explains where the cost really sits, how to estimate it yourself, and which question decides whether it is worth it.

Looking for the bigger picture first? Read AI for your SME. Want to know what determines the price? Read on below.

Three kinds of cost ​

An AI solution costs you in three places. Anyone who names only one is forgetting something.

  • The tool or subscription. An off-the-shelf AI tool often costs a few tens of euros per month per user. This is a recurring cost, like any other software.
  • The setup. One-off work to fit the tool to your situation: opening up your data, configuring the task, testing and tuning until it is right. This cost depends strongly on how complex the task is.
  • The upkeep. Small adjustments when your process changes or when new exceptions appear. Count on little here, but not on nothing.

The tool is usually the smallest item. The setup makes the difference between a cheap and an expensive solution.

The only question that matters ​

The question is not "what does the tool cost". The question is: does the time saved on one task cover that cost?

Take a concrete task. How often per week does it recur, how many minutes does it take each time. Work it out over a year. Ten minutes, five times a day, is more than 200 hours a year. Put the hourly cost next to that and you know what the task costs you today.

Compare that with the price of the solution. If it only pays back after three years, leave it. If it earns itself back in a few months, you have a strong candidate. That calculation is more honest than any price list.

What pushes the price up or down ​

Two businesses with the same request rarely get the same price. Four things explain the difference:

  • The number of exceptions. A task with a fixed pattern is cheap to set up. If it is full of special cases, each of those takes time to capture.
  • Links with your systems. If the solution works on its own, it is simple. If it has to talk in and out of your accounting, your CRM or your web shop, the cost rises.
  • The state of your data. Clean, findable data saves work. Tidying up scattered or messy data first costs time, and therefore money.
  • How often it changes. A stable process you set up once. If it changes every month, count on ongoing adjustment.

Want to start cheaply? Choose a task with few exceptions, few links and clean data.

Pay for the result, not for the promise ​

A large budget up front for something unproven is the biggest risk. It can be the other way around: start small, measure, and pay only as it works.

That is how we work. We agree a measurable goal up front, in hours saved and errors avoided. If that goal is not met, you get your money back. That way the risk sits in the right place, and you pay for what works rather than for a quote.

What to watch for in a quote ​

Ask every provider the same questions, and compare the answers:

  1. What is one-off and what is recurring in this price?
  2. What measurable result stands against it, and how is it measured?
  3. What happens if the goal is not met?
  4. What does an adjustment later cost, if my process changes?

A provider who answers these clearly knows what they are talking about. One who names only a total amount is hiding where the cost sits.

Frequently asked questions ​

Is there a minimum budget to start? Less than you think. A first, well-defined task costs little to set up. The question is not whether you have the budget, but whether the time saved on that task covers the cost.

Do I pay per month or once? Usually both. The tool is a recurring cost, the setup is one-off. Always ask for the split, so you know the yearly cost and not just the starting amount.

Am I eligible for a subsidy? Support schemes change regularly, and not every kind of AI work qualifies. Get informed in advance and never count a subsidy into your payback time before it is confirmed.

What if it does not deliver what was promised? Agree that up front. A clear goal and an agreement on what happens if it is not met keeps a test a test. Want to check without commitment what a task would cost? Submit your question here.

Further reading ​

Coronis tells you the odds and the price up front, and agrees a measurable goal you can hold it to. Pay for what works, not for a promise.

Written with AI, read and approved by a human. Here is how that works.