Why measurement matters

The problem is the estimate, not the person estimating.

When people judge liquid by sight, the answer can depend on the container and the observer. Research in other settings helps explain why CountMesh is being built around a confirmed weight instead of a visual fraction.

The evidence

What researchers actually tested.

These studies concern visual judgments of liquid in cups or glasses. They are relevant to the counting problem, but none tested restaurant staff inventorying liquor bottles.

RESIDUAL LIQUID / 2022

Visual estimates varied, even among experienced nurses.

Michelsen and colleagues asked nurses and other participants to estimate liquid left in, or removed from, familiar containers. They compared estimates with volumes measured using a digital scale. For the nurse group, mean estimates fell within ±10% of the actual amount in only half of the tested conditions.

Limit: Those are group means across conditions—not the percentage of nurses who were accurate. The containers were clinical cups and glasses, not liquor bottles.

Read the study

RESIDUAL LIQUID / 2016

Experience did not erase the spread of estimates.

In Tattersall's study, 123 nurses and care staff estimated fluid consumed from what remained in six everyday containers. Only 25% of individual estimates were within the study's ±10% margin. The study found no statistically significant difference by length of time working in healthcare.

Limit: This was a clinical fluid-intake task, and finding no significant difference does not prove experience has no effect in every setting.

Read the study (PDF)

CONTAINER SHAPE / 2015

Shape changes how volume looks.

Pechey and colleagues asked 360 adults in an online experiment to match perceived wine volumes across glasses with different dimensions. The volume that looked equivalent changed with glass size and shape.

Limit: Participants compared images of wine glasses; they did not handle bottles or perform an inventory count.

Read the study

What it means for the count

Give different counters the same measured input.

CountMesh is being designed to read a digital scale's whole-gram display, have a person confirm the weight, and use bottle-specific reference weights to estimate what remains. That changes the counter's normal task from judging a fraction by eye to checking a measurement.

A weight alone is not an inventory quantity: the right product, bottle size, and empty and full references still matter. These studies do not measure CountMesh's accuracy, speed, or business impact. Those claims will require testing the actual workflow.

See how the approach works

The honest conclusion

Visual liquid-volume estimates can vary, and container shape can influence them. A confirmed weight offers a common starting point—not a promise of perfect inventory.

Help us test it in the field

What does counting look like at your location?

We would like to understand where visual estimates cause disagreement and what a measured workflow would need to fit.

Talk to us