Skip to content

What does analytics measure, and what does it not?

Play, valid play, contact, and impact, the validity rule, measured, derived, and estimated, the planned time, and the limits of the measurement.

8 minute readAnalyticsUpdated: 10/25/2026

On this page

Before you start

For
Everyone who wants to know or prove what ran, from the shop checking the display in its window to the advertiser receiving a proof.
Prerequisite
The permission Run ad hoc queries, held by Owner and Manager by default.
Where in the portal
Analytics

The program says what should run. Analytics says what was visible: Every play is measured on the display, judged by a rule with a stated version and, where an audience is on file, converted into contacts. Every number also says whether it is measured, derived, or estimated.

Four stages, each building on the previous one

The smallest unit is the play: one piece of content, once, on one display, from start to finish. Content is a scene, a media item, the creative of an ad campaign, or the filler of an ad break. If a scene, a media item, or a filler runs longer, the display reports it in sections of at most five minutes, and each section counts as a play of its own. That way, even content that runs all day appears in analytics after a few minutes, and if the display software stops, only the last minutes of it are missing. The airtime stays the same. Creatives of an ad campaign are never split. The display reports what it actually showed, not what was planned: If it falls back to a different scene, that scene is in the measurement. It also reports the circumstances of each play: whether the first frame appeared or an error left the area empty, how long the display software was covered, which file it showed, and how far its clock deviated from the server’s. Separately, it reports whether the TV was on or on standby and when it was recording at all.

On the server, the validity rule judges every play. A valid play has met all conditions of the rule, and every other play carries the first reason it failed. Valid plays and the audience of a location produce contacts: modeled opportunities for a person to see a play. The fourth stage, the impact as a comparison of sales, does not exist in the portal yet.

A play is measured on the display, the validity rule in version 1 turns it into valid plays, the contact model in version 1 turns those into contacts, and the impact, as a comparison of sales, follows later

Each stage builds on the previous one and does not change it. A new version of the rule recalculates the stages above it, while the measured plays stay as they were reported.

A creative that ran three times for ten seconds makes three plays and thirty seconds of airtime. Airtime is the sum of the duration of valid plays. If two plays overlap at their boundary, the shared time counts only once, in analytics as in the proof. Anyone who sells ad space counts in plays, and anyone who assesses operations counts in airtime and availability: the share of the planned time in which the display was visibly running.

The validity rule, version 1

The rule checks in order. The first verdict that applies stands.

Verdict When
playback error The first frame did not appear, or an error left the area empty.
wrong content For a creative: The display did not show the approved file, the creative is not approved, or it does not belong to the creatives of the booking.
panel off The TV was on standby for more than a tenth of the duration.
not visible The display software was not in the foreground for more than a tenth of the duration, for example under a TV menu, or it was dimmed almost to black (software dimming to five percent or less).
zone too small For a creative: The area was smaller than the minimum share of the booking, by default a quarter of the display area.
too short For a creative: The play ran shorter than 90 percent of the booked spot length. A video must also have run to its end or have filled the slot.
outside the booking For a creative: The start was outside the period or dayparts of the booking.
not planned For a creative: The planning did not provide for another play of the booking on this display in this hour. At most the planned plays and two more count, the earliest first. Without planning in the hour, none counts, for example on a display outside the targeting.
clock unreliable The display had no time source, its clock deviated by more than five seconds, or two of its plays overlap by more than two seconds. A display never shows two things at once, so the timestamps contradict each other.
valid None of the conditions triggered.

For own content, that is scenes, media items, and fillers, the rule checks only for playback errors, panel, visibility, and clock. Until the verdict is in, a play that has arrived appears as being checked, usually for a few minutes.

Only what the planning provided for is counted. For every booking, it works out how many plays are meant to run on each display in each hour, and it freezes that figure as soon as the hour begins. A new booking therefore counts from the next full hour after it is booked.

If you change a running booking or a creative, for example the dayparts, the minimum share of the area, or the mandatory text, the change applies to plays after that. A play that already has a verdict stays judged by the conditions under which it ran. Only what the display itself measures can still change such a verdict: playback, panel, visibility, and clock, for example when the state of the TV is reported later.

If a TV does not report its state because it answers neither over HDMI-CEC nor over DDC/CI, its plays count as valid. They are marked, though, as without panel confirmation, in the day view as well as in the proof: Display control over HDMI.

Measured, derived, estimated

Every number in analytics carries its class. A filled circle with M stands for measured, a ring with D for derived, a dashed ring with E for estimated, and estimated numbers also carry a ≈.

Class What it means Examples
M measured collected on the display Plays, valid plays, airtime, visible time, reporting gap, reported late, playback errors, reached locations
D derived calculated from measured numbers and the plan Planned time, availability, planned plays, fulfillment, pacing, failure rate, share of display, utilization
E estimated from a model with a named source Contacts, contact value

Measured and estimated numbers never appear in one sum. Anyone who quotes contacts to an advertiser also states the grade of their data: Set up ad space and audience.

Planned time: when a display should show content

The planned time is the time in which a display should show content according to opening hours and program, without off-air time. A shop that is closed at night has no planned time at night and therefore no gap either. Without opening hours on file, analytics calculates the planned time from the program alone. You maintain the opening hours at the location: Create locations and assign displays to them.

The calculation uses the program that applied at the time, not today’s. Every change to a program, an override, or a playlist freezes the plan for the next 48 hours, and past hours are never recalculated. The opening hours, in contrast, follow the current state of the location.

Where measurement ends

  • The screenshot shows the output of the display software, not the panel. Videos that a device plays on a layer of its own are missing from it, and a TV on standby looks on the screenshot like one that is showing content. What the panel did is stated by the measurement of the panel.
  • The display signs what it reports with a key that is stored as a file on the device. The signature proves that a batch came from this device and was not changed afterward. It does not protect against a device that someone has modified. Web displays have no device key, so their reports appear as unsigned.
  • Without CEC or DDC/CI, there is no panel state. Plays on such displays count and appear as without panel confirmation.
  • Contacts are a model. They are calculated from visitors per hour, dwell time, and placement, never from people in front of the display. Displays have neither a camera nor sensors that detect people.
  • The visitor figures are declared by the location. Grades A and B also follow from a count, a survey kit, or till receipts that the location enters itself. Szenera does not verify these figures so far: Set up ad space and audience.
  • Measurement starts with the version of the display software that reports plays individually. Keep your devices up to date: Updates.

What follows

  • Where nothing was reported, there is never a zero but a gap: Why is a gap not a zero?
  • All days and hours are in the time zone of your organization. The header of Analytics names it with Times in ….
  • Individual plays are kept for 13 months, hourly values for 25 months, and daily values indefinitely. Szenera keeps proofs for ten years.
  • If a number differs from your expectation, that is a finding about reality: Check a display’s day.

Found what you needed? Then carry on in the portal — or set up an organization first.

Go to the portalTry it free

Cookies & analytics

We only use technically necessary storage to run this site. With your consent, we additionally measure anonymized usage — it helps us make the site better. Privacy policy

You decide what this site may do. Your choice applies until you change it here.

Necessaryalways on

Language preference and your cookie decision. The site does not work without them — always on.

Anonymized usage statistics. Loaded only after your consent and revocable here at any time.