Resources · Measurement

How measurement works.

Measured exposure is the unit the whole platform is built on. This is the methodology behind it: what counts, how it is meant to be captured, and how rendered time becomes a billable total. It is also an honest account of what is built today and what is still being built.

The principles

Three rules the measurement model is built on.

Before any of the mechanics, the model rests on a few commitments about what measured exposure is, and what it deliberately is not.

Read on the serving side

Delivery is meant to be measured where the ad is served, not claimed by the creator and not inferred from a screenshot after the fact. The number a brand pays against is designed to be one nobody has to take on trust.

Rendered time is the unit

The meter is built to run while the creative is actually on the screen and to stop when it is not. A dropped stream, an off-screen card, a paused placement: those are designed to be time that is not counted and not billed.

Privacy by construction

Measurement here is about what the system put on the screen, not about who watched. We do not track viewers or build audience profiles, so there is no personal profile behind a measured number.

The methodology

From on screen to a billable total.

The model works in three moves: define what counts as rendered, capture it off the running stream, and reduce it to the exposure a flight delivered. The serving log is the artifact that falls out of it, a second-by-second record meant to be read rather than trusted.

1

Define what counts as rendered

The model sets a precise rule for when an ad is considered on screen and therefore billable. This definition, and how it turns seconds into a total, is the part that is built.

2

Capture it off the running stream

The live capture path reads what is actually on the broadcast, second by second, and feeds it into the model. This path is still being built and carries privacy and platform terms we are working through first.

3

Reduce it to a billable total

Rendered seconds are reduced to the exposure a flight delivered, and the flight is designed to close with a serving log a brand can read line by line.

Serving logExample

How a lower-third is designed to meter, second by second:

20:04:11Lower-third on screenrendered
20:04:12Lower-third on screenrendered
20:04:13Lower-third on screenrendered
20:04:14Stream dropped, serving pausednot billed
20:04:19Lower-third on screenrendered
You pay forrendered seconds only

Illustration of the metering model, not a real campaign. It shows how measurement is designed to work, not results.

Where it stands

What is built today, and what is still being built.

We are building this in the open and being precise about where we are, because a measurement claim is only worth anything if it is honest about its own status.

Honest status

The model is built. The live meter is still being built.

The metering model, how rendered time is defined, reduced, and turned into a billable total, is built. The live capture path that streams every second off a running broadcast into that model is still being built and carries privacy and platform terms we are working through first. So the honest claim today is about the design, not a finished, fully running meter: this page describes what measurement will count and how it is built to count it, and we will say plainly when the live meter is fully running.

A measurement claim is only worth something if it is honest about its own status.

So we describe the model as designed, and we tell you exactly what is running today.