MMOOZZGG.com

AI solutions map

AI ROI calculation: how we count the effect of AI in money

5 min read2 sources

We make the AI ROI calculation in money before the start, with a formula you can check. According to the MIT NANDA report of 2025, 95% of organizations got no measurable return from generative AI. So every solution on the map gets a calculation and a metric that are checked after launch.

On this page · 5 sections

What the effect is made of

The effect has two parts: savings and growth. In the calculation before the start we take only savings, because they can be counted from the company’s records.

Monthly savings = number of employees on routine x salary x 1.3 x share of work AI takes over. The factor 1.3 covers taxes and contributions on top of take-home pay.

The share of work does not mean “we will replace a person”; it is part of their day: answers to routine questions, moving requests, reconciliations, reports. We estimate it for each process separately.

Almost the whole result depends on the share of work: with the same headcount and salary, savings rise and fall with it. That is why we estimate it by process instead of taking an average.

Growth, such as more requests handled or a reply to a customer in minutes instead of the next day, is not part of the calculation before the start. It shows after launch and comes on top of the calculated savings.

What we subtract from savings

From savings we subtract everything the company will pay on an ongoing basis. Support comes first: from $480 a month, and you can cancel in any month. Without it, in six months AI works by outdated rules.

Second, we subtract the share that does not go to AI. A common mistake is counting all of an employee’s work, although AI takes the routine part and the hard cases stay with people.

Third, we count your people’s time: someone must review disputed cases and accept fixes. If there is no such person, the savings move months later.

The one-off rollout price is subtracted in the yearly calculation. A pilot costs from $1,499, an AI department from $4,999, a turnkey business from $18,000.

Right below this section is the payback calculator, the same as on the home page. Enter your own numbers: support is already subtracted in it.

Payback calculator

How much your AI department saves.

What we calculate
Plan

An estimate. Salary with payroll taxes ×1.3. Support at $480 a month is already subtracted.

Savings per year
$22,001
Per month
$2,250
Payback
3 mo.
Check it on my business

Payback period

Payback period = one-off rollout price / monthly savings after support. Savings in the first year = monthly savings x 12 minus the one-off price.

An example with the calculator settings: employees on routine 5, salary $950, share of work for AI 60%, support $480 subtracted. Monthly savings: $3,220.

For an AI department at $4,999 this is $33,600 in the first year and a payback of 2 months. The numbers change with the inputs: with a smaller share of routine, the period grows.

Why we calculate conservatively and which calculation errors we see most often is covered in the article on AI department payback.

Why the metric is written into the contract

The metric from the calculation goes into the contract: what it was, what it should become and how to measure it. For example, the share of requests an AI agent processed without a person, or the time to reply to a customer.

This way the result is checked by a number, not by the fact of launch. According to MIT NANDA 2025, only 5% of custom AI tools reach real use, and a launched system says nothing about money yet.

We choose the metric together with the process owner, not with the IT department: the owner is responsible for the number and knows how to measure it.

After launch we take the metric report and put the real share of work into the same formula. If the number is below the calculation, the cause is visible: people do not give requests to AI, the rules are outdated, or the process has changed.

How to choose a metric before the start is covered separately.

Where there will be no effect

1. There is less routine than one employee’s worth. Then AI solves a quality task, such as answers at night and at weekends, not a savings task.

2. The process is not described. AI has nothing to learn from, and the first weeks go into describing the process instead of results.

3. There is nobody who decides on fixes. AI keeps making mistakes in the same places.

4. The records are not in order: stock, prices or customer history live in people’s heads and scattered spreadsheets. First put the records in order, then AI.

5. The calculation assumes replacing people completely. AI takes part of an employee’s day, and a calculation on the whole salary gives a number that the report will not confirm.

In these cases the map shows “nothing” or process preparation, and we say so before the contract, not after.

Questions and answers

How fast does AI pay back?

It depends on the share of routine and the price of the format. In the calculator example with the default settings, an AI department for $4,999 pays back in 2 months, with monthly savings of $3,220 after support is subtracted. If there is less routine than one employee’s worth, there will be no payback, and we will say so in the review.

Sources and fact-checking

Checked by the editors on 13 October 2026.

  1. MIT NANDA: The GenAI Divide, State of AI in Business 2025 (PDF) (mlq.ai)
  2. Virtualization Review: MIT report finds most AI business investments fail, reveals GenAI Divide (virtualizationreview.com)

Found a mistake? Tell us

Publications

Contact us