On 15 September 2026 we told the market, in current report ESPI 11/2026, that we were ending regular publication of our case volumes and our MPC, the minutes of processed conversations. We did it because those numbers had stopped describing the economics of the business, and an indicator that no longer describes the business is worse than no indicator at all.

For years, minutes were our headline operating metric, and investors followed them closely. This piece explains why we dropped them, what the decision says about the direction of the market, and what we intend to report instead. The short version: when clients pay for effectiveness rather than talk time, a minute in one process can be worth several times a minute in another, and adding them up tells you very little.

Why minutes made sense for so long#

Minutes were not a mistake. When automated telephone contact was young, the industry inherited its units from telephony and from call-centre seat economics: lines, channels, minutes. Most clients bought voice automation as capacity, to replace or supplement agent hours, and a minute of AI conversation was a natural like-for-like comparison with a minute of human conversation. For a growing company, rising minutes were also an honest signal that clients were using the service more.

The unit stops working once the service changes. We do not sell capacity; we run processes and answer for their results. Once the work includes number strategy, retry timing, SMS sequences, data exchange, daily reports and continuous optimisation, the conversation becomes one part of a larger job, and its length becomes a poor guide to the size or value of that job. In a database with low reach and frequent hang-ups, a large amount of work produces very few minutes.

What changed in the numbers#

Minutes and cases used to move together closely enough for minutes to act as a proxy for revenue. In 2024 we handled 8.2 million cases and 5.0 million minutes; in 2025, 11.61 million cases (up 45.7%) and 7.18 million minutes. The two series rose at a similar pace, and most of our contracts were priced in a way that made talk time a reasonable guide to activity.

Since the start of 2026, minutes have been falling faster than cases. There are three reasons, and none of them is a problem:

  • Shorter conversations. We keep redesigning dialogues to reach the commitment sooner. In banking, cases ending with a promise to pay averaged 116.5 seconds of handling time (historical). A shorter call that ends in a commitment is a better call.
  • A changing campaign mix. Different processes have very different conversation lengths. A shift from lead confirmation towards early receivables monitoring changes total minutes without saying anything about performance.
  • Foreign processes. Our share of contact-process revenue from outside Poland rose from 6.78% in Q4 2025 to 21.95% in Q3 2026 and 25.27% in October 2026 (excluding implementation fees, additional services and one-offs). Those processes have a different profile of length and pricing.

In the third quarter of 2026, total conversation time fell considerably faster than the corresponding traffic revenue, and the average revenue per unit of time changed. Conversation volume had stopped being an adequate description of the economic weight of what we do. Continuing to publish it as a headline would have invited exactly the wrong conclusions.

Equal minutes, different value#

The deeper reason is about billing. When a client pays per minute, every minute is worth the same to the supplier, and minutes are a fair measure of the supplier's revenue. When a client pays per case or account, with a performance layer that rises or falls with agreed KPIs, the link breaks. A two-minute call that secures a payment date for a debtor about to slide into the next bucket can be worth far more than a six-minute call that ends in “I'll think about it”.

Some of the most valuable cases involve no minutes at all. A customer sees a missed call from their bank and pays the same day. In a minute-based view that case is invisible; in a case-based view with the bank's own business status attached, it is a result. A metric that cannot see the result is not a performance metric.

If we are paid by the minute, we are not playing towards the same goal as our client. – Dawid Wójcicki, Founder & CEO of Vee

Our billing models still coexist. New agreements are no longer billed by the minute; some existing contracts still are, and the right model depends on the process, the terms of cooperation and whether results can be measured reliably in the client's systems. But the direction is clear, and outcome-linked models are gaining weight. A KPI Incentive model, for example, is being prepared with a client outside Poland. Reporting had to follow the business, not the other way round.

We started moving away from minutes earlier#

The decision did not come from nowhere. In May 2025, when we published our April figures, we added case counts alongside minutes for the first time, explicitly to measure work regardless of conversation length. The trade press noted it at the time: more cases, fewer minutes. That was the first public signal that talk time was becoming the wrong unit.

The argument is older still. In December 2022, in an interview with NowyMarketing on AI in e-commerce, our CEO argued that the relevant figure is the cost of a single success, not the price of a call: a cheap solution that converts a small fraction of contacts can cost far more per success than a dearer one that converts many times as often. Three years later, the same logic applies to how a company like ours should describe itself to investors.

Volume and economics have already separated#

Our public results show the separation plainly. In the first nine months of 2026, consolidated revenue rose 12.6% year on year while operating profit rose 177.4%, and the operating margin went from 19.8% to 48.9%. In the third quarter alone, revenue was 1.4% lower than a year earlier while operating profit rose 142.7%. Those figures come from our Q3 2026 quarterly report. They describe a business whose economics are driven by the structure of its processes, its billing models and its use of resources, not by how many minutes it talks.

An investor reading minutes alone in 2026 would have drawn the wrong conclusion about the direction of the company. An investor reading margins and the shift in revenue geography would have seen a business becoming more profitable and more international at the same time. When two honest numbers point in different directions, the one that does not describe value has to go, however familiar it is and however long the market has followed it.

The market is arriving at the same place#

We are not alone in moving the unit from time to result. In January 2026 Gartner forecast that the cost per resolution of generative AI in customer service will exceed $3 by 2030, above the cost of many offshore human agents. That is a forecast, not a finding, but its framing is telling: the analyst's unit is the resolution, not the minute and not the API call. As automation becomes cheap to start but not free to run, the question every buyer will ask is what a result costs.

For a contact operator, that question cannot be answered with minutes. It needs reach, continuity, key information and business outcome, measured on the same buckets and portfolios as a comparator, in the client's systems. Those are the numbers that tell a client, or an investor, whether the work is creating value.

What we will report instead#

Our work on a new operating-reporting standard focuses on three complementary dimensions:

  1. Economic value: revenue and the structure of the processes we run.
  2. Operational effectiveness: understood through measurable results rather than activity.
  3. Process economics: the relationship between the value of the service, the effort needed to deliver it and the results achieved.

Building that set of indicators means fixing definitions, data sources, methodology and rules for comparing periods. We would rather take the time to get it right than publish a new headline number that has to be explained away a year later. Investor-relations materials in eleven languages are scheduled from the Q4 2026 report, due in February 2027, with the official Polish documents remaining the binding versions.

What this means for buyers of customer contact#

The lesson is not specific to listed companies. If your supplier's reports lead with minutes handled, they describe the supplier's capacity, not your results. If its invoice is driven by minutes, its incentive is to keep conversations long. Neither tells you whether more customers paid, attended, renewed or bought.

There is a practical side too. Minute-based reports are easy to produce and easy to audit, which is why they persist. Case-based reporting with business statuses from your own systems takes more work to set up: agreed definitions, a daily data exchange, a measurement window and a comparator. That work is the price of knowing whether the contact is paying for itself. It is also what turns the weekly review from a debate about volumes into a decision about what to change next.

  • Ask which unit your supplier reports first, and which unit it is paid in.
  • Ask for reach measured per case, not per attempt, and for results measured in your own systems.
  • Ask how the supplier treats cases that are resolved without a conversation.
  • Ask whether its fee can rise or fall with the KPIs you actually manage.

We stopped publishing minutes because they had stopped telling the truth about our business. If they are still the main number in your contact reporting, it is worth asking whether they are telling the truth about yours.

Follow The First Second

Occasional issues on the mechanics and economics of AI-powered customer contact. No noise, no vendor news.

Follow via RSS