Screen uptime: the metric underneath every impression
Every impression estimate for a screen network rests on an assumption that the screen was powered, connected and playing. Uptime is the measurement of that assumption, and it is the least glamorous and most load-bearing number in the channel.
A network that reports impressions without reporting uptime is reporting a model output.
What causes downtime?
Power interruptions, network connectivity failures, hardware faults, and stores switching screens off. The last is specific to independent retail and it is the one a brand is least likely to anticipate.
A store owner who finds a screen noisy, bright or in the way may simply unplug it, and unless the network detects that, the screen is reported as delivering.
How is uptime measured?
By the player reporting in, and by playout logs showing content actually rendered. Heartbeat monitoring shows the device is alive; playout logs show content played, which is the stronger evidence.
Ask which the network uses. A device that is powered but showing a blank screen or a frozen frame can report healthy on a heartbeat.
Why does this matter more in independent retail?
Because the estate is fragmented and the screens are in premises the network does not control. A chain can mandate that screens stay on; an independent operator makes their own decision every day.
That is not a criticism of the channel, it is a structural property of it, and a network that monitors closely and reports honestly is managing it. One that does not is reporting an estate it is not observing.
What should a contract say?
What uptime is measured, how, what threshold triggers remediation, and what happens to delivery that did not occur. Make-goods for undelivered plays are a normal commercial arrangement and they depend on uptime being measured.
Without a measurement, under-delivery is invisible and the make-good clause is decorative.
How should uptime appear in reporting?
As a percentage by store group and period, alongside delivered plays. Those two together tell a buyer both what ran and what did not.
A network that reports a single network-wide uptime figure is hiding the distribution. A small number of screens dark for a long period is a different problem from all screens dark occasionally, and only a distribution shows which.
What should a buyer do with a low figure?
Ask for the distribution, the causes and the remediation timeline, and check whether the affected stores were ones you were paying for. Uptime is not uniformly valuable: a dark screen in a high-traffic store matters more than one in a marginal store.
Does uptime affect measurement of effect?
Yes, and it is frequently ignored in lift analysis. A test store whose screen was dark for a meaningful part of the flight is a control store with extra steps, and including it dilutes the measured effect.
Compliance data belongs in every measurement of this channel for exactly that reason.
Frequently asked questions
What uptime should be expected?
High, and networks should be able to state their own figure and their method. Be sceptical of a number quoted without a definition, because uptime measured by heartbeat and uptime measured by verified playout are different measurements.
Who is responsible when a store switches a screen off?
Commercially, the network, since the advertiser bought delivery. Practically, it is a field relationship question, and a network with local field presence resolves it faster than one relying on remote monitoring alone.
Can I get store-level uptime?
Ask, because some networks provide it and some report only aggregates. Store-level data is what allows a measurement analysis to exclude non-compliant stores properly.
Does downtime get credited?
That depends on the contract. Make-goods, extensions or credits are all used, and the mechanism should be agreed before the flight rather than negotiated afterwards.
How quickly should a fault be detected?
Within a day for a monitored network, because a fault undetected for a week is a week of delivery lost. Ask what the monitoring cadence is and what triggers an alert.
Is a frozen screen counted as up?
It should not be, and a heartbeat-only system may count it as such. This is the specific reason to ask about verified playout rather than device status.
Does this apply to static signage too?
The equivalent question for print is whether the material was actually posted and remained in place, which is an audit rather than a monitoring problem. Verification practice in out-of-home is discussed by bodies such as Geopath.
What about scheduled downtime?
Store closing hours are legitimate and should be excluded from the calculation by definition rather than counted as downtime. Confirm how the network handles hours, because it changes the denominator.
What is the single question worth asking?
Is uptime measured from device heartbeat or from verified playout. The answer tells you how much the impression figures are worth.
### What does a good uptime report look like?
A distribution rather than an average: how many stores were live for the whole flight, how many had partial outages, and how many were dark for extended periods. Three buckets tell a buyer far more than a single percentage.
It should also identify the stores, so that delivery can be reconciled and measurement can exclude them. An aggregate figure supports neither.
Should uptime affect what is paid?
It should feed the make-good mechanism, which is why the definition matters commercially as well as analytically. A contract referencing delivered plays handles it automatically; one referencing a schedule does not.
### How should uptime feed into planning?
By discounting expected delivery. A plan built on nominal capacity without an uptime assumption will overstate delivery by whatever the estate's real figure is.
Using the network's historical distribution rather than its headline average produces a more realistic plan, and it makes any shortfall a smaller surprise.
Should uptime targets be contractual?
A target with a remedy attached is meaningful; one without is an aspiration. If uptime is in the contract, the make-good mechanism should reference it.
Does seasonality affect uptime?
It can, through power interruptions, store refits and holiday closures. Asking for uptime by month rather than annually shows whether the estate has a seasonal pattern.
Considering NRS Digital Media? Start at nrsdigitalmedia.com, read the media deck, or contact the team.