AI and profit­ability: how to assess an investment

Separate the business benefit from the tool’s promise

AI may help a business reduce work, improve service or process information. Using a tool does not automatically improve profitability. Benefits must exceed the cost of implementation, use, checking and maintenance.

Start with an identifiable problem: which task consumes time or creates costs, how is it done now, and what would change? This article discusses investment assessment, not a specific tool purchase or Revla service prices.

Where can the benefit come from?

1. Releasing working time

Assisting repetitive tasks may save time. Measure the complete process, including instructions, checking and corrections. A faster draft does not necessarily mean a faster finished task.

Released time is not automatically a cash saving. It might increase capacity or support better service. Financial value depends on whether an actual cost falls or the time enables additional profitable work.

2. Improving customer acquisition

AI can help draft content variants or organise material. Sales growth cannot be inferred from content volume. Monitor whether suitable enquiries arrive and become sales.

An advertising platform’s automation optimises towards the objectives it receives. A target of cheap form submissions does not tell it which customer produces the best contribution. Sales feedback and accurate measurement remain essential.

3. Processing information more effectively

AI can help classify feedback or propose interpretations of material. Check suggestions against the original information. A convincing explanation is not proof of a correct conclusion.

Value arises when an observation leads to a useful decision or change. Producing a report faster does not by itself establish better business performance.

How do you calculate ROI?

Compare benefits and costs over the same period. Count realised cost savings or contribution from additional sales as financial benefits, rather than revenue growth alone.

ROI = (financial benefit − total cost) / total cost × 100%

Include implementation, integrations, training, internal time, checking and maintenance alongside the tool cost. Avoid double-counting a benefit as both saved time and the sales that time enables unless you can justify the separation.

Prepare at least a conservative estimate and a target scenario. If the investment only works under optimistic assumptions, investigate the largest uncertainty before a broad rollout.

Example: saved time and recurring cost

The following numbers are hypothetical assumptions, not service prices or client results.

Suppose a trial releases five hours a month after checking time is included. At an assumed value of €100 per hour, that represents €500 of monthly capacity. It does not yet establish a €500 cash saving.

With an annual tool cost of €2,000, the recurring monthly cost is about €167. The assumed benefit less the tool cost is about €333 per month before other costs. An annual subscription alone does not establish a five-year payback period.

If implementation also required a one-off €1,000, recurring costs were paid evenly each month and the full assumed benefit became financial value, simplified payback would be about three months: 1,000 / (500 − 2,000 / 12). If the time does not create financial value, or other costs arise, the calculation changes.

The example distinguishes a one-off investment from a recurring expense. Use your own data for a real decision.

What should a limited trial measure?

Record a baseline before changing anything and compare similar tasks under similar conditions. Monitor total time, output quality and corrections. Marketing work also needs enquiry and sales quality measures.

Decide beforehand which results justify continuing, changing or stopping the trial. Its duration depends on the task and number of observations. A generic promise of results within a particular number of months cannot replace that plan.

When should you avoid investing yet?

Do not buy a solution if you cannot name the problem or its current cost. Check whether simpler processes or ordinary rule-based automation would be sufficient.

If data permissions are unclear or results cannot be checked, limit the trial to suitable material. If benefits are small relative to total costs, change or abandon the project. Payback within 18 months alone does not prove an investment is worthwhile: risk, cash flow and alternatives also matter.

Where should you start?

Choose one problem, measure the current task and estimate benefits conservatively before selecting a tool. Expand when the value and quality have been demonstrated in practice.

If the relationship between marketing investment and results is unclear, contact us. We can discuss your current situation and assess what needs investigating first. Our current offering is described on the consulting service page.