---
title: "How we measure consumption at a dispenser"
description: "The methodology behind the figures we publish: what is metered, over what period, what is excluded and what the limits are. Without it, any consumption figure is just an assertion."
category: studii-de-caz
slug: cum-masuram-consumul-la-un-dozator
date: 2026-10-27
---

A consumption figure without a methodology is not a result, it is an assertion. Before any case
study, we publish how we measure — what is metered, over what period, what is excluded and where
the limits are. Anyone who wants to challenge one of our figures should be able to attack the
method, not just the number.

## What is metered

The dispenser records the **volume served**, tied to the serving session. Each serving is
associated with an account in the Unikat app, because access works on a subscription.

From a measurement point of view, that means two good things and one limitation:

- **good:** the volume is measured by the machine, not estimated from the number of containers sold
- **good:** each serving has an exact timestamp, so the distribution across hours can be reconstructed
- **limiting:** what is measured is what was served, not what was drunk

The second point is important and we do not dodge it: between "served" and "consumed" there is a
difference no dispenser can measure. Water left in a bottle at the end of a session was served but
not necessarily drunk.

## The reporting unit

We report **litres per training session**, not litres per member and not litres per day.

The reason: a gym's membership count changes monthly and includes inactive accounts. The number of
sessions is closer to physical reality — people who came in and trained.

To obtain the figure, we divide the total volume served in a period by the number of sessions in
the same period, at the same location.

## The period

A water consumption measurement needs a stated period, for two reasons:

1. **Seasonality.** Consumption rises in hot weather. A figure from August and one from February
   describe different things about the same gym.
2. **Novelty effect.** The first weeks after installation have a pattern of their own: people try
   the machine. They are not representative of the steady state.

That is why every figure we publish comes with the month and the year. A figure without a period
does not get published.

## What we exclude

- **The commissioning period**, until the pattern of use stabilises.
- **Test servings** from maintenance and technical checks.
- **Days when the serving point was unavailable**, from both the numerator and the denominator of
  the ratio.

## The limits we declare

**We measure serving, not consumption.** Explained above.

**An average hides the distribution.** An average consumption per session says nothing about how
many people drink a lot and how many drink nothing. Two gyms with the same average can have
completely different patterns.

**The result does not transfer automatically.** A figure obtained in a gym does not apply to an
office space, where the consumption pattern is different — more spread across the day, without
exercise-driven peaks.

**The number of locations matters.** An average across five locations is more robust than one
across a single location, but it remains an average across five locations, not a general law.

## What we do not do

We do not estimate. If we do not have the measurement for a location and a period, we do not
publish a figure for them — not "approximately", not "around", not extrapolated from another
location.

We do not attribute health effects to measured consumption. Volume served is a logistical
quantity.

Water contributes to the maintenance of normal physical and cognitive functions, at a daily intake
of at least 2 litres from all sources — that is
[the only claim authorised in the EU](/en/stiai-ca/mentiunea-de-sanatate-autorizata-pentru-apa/).

## Key points

- We meter volume served, tied to the session, through the app.
- We report litres per session, not per member.
- Every figure comes with the month, the year and the number of locations.
- We exclude commissioning, maintenance test servings and days of unavailability.
- Served does not mean drunk, and an average does not describe the distribution. We say so every time.

## Sources

- Own metering data, Unikat Flow system
- Regulation (EU) No 432/2012 — EUR-Lex, CELEX 32012R0432
- Regulation (EU) No 1169/2011 — EUR-Lex, CELEX 32011R1169

---

Source: https://unikatflow.com/en/studii-de-caz/cum-masuram-consumul-la-un-dozator/