- Published
- Updated
- Updated
- Reading time
- 10 min read
Digital Signage Proof of Play: 10-Point Buyer Checklist
Digital signage proof of play checklist: clock sync, playback logs, brightness drift and privacy—verify auditability before you buy. Get the spec sheet.

On this page
Auditable proof of play is not a dashboard feature you buy — it is a three-record chain that has to agree: the CMS publish log (what was scheduled and sent), the player playback log (what the device actually rendered, timestamped on the device), and, where your contract demands it, panel-side confirmation (the display's own brightness and power state at the moment of play). The audit survives only if all three hold, and the weakest link is almost always the player's behaviour during a power cut or a WAN outage — not the CMS report your vendor demos. Note the boundary before you spend anything: as Videri stated in its ISE 2026 position (videri.com), proof of play shows that an ad ran, and when and where — it does not show that anyone saw it or that it changed a purchase. Separate the delivery record you must defend contractually from any audience-measurement layer you buy separately.
The gap is not theoretical. Kitcast's State of Digital Signage (July 2026, N=515 survey plus telemetry from 10,520 workspaces), as reported by CrownTV, found that 76.5% of operators never formally measure ROI. A network that cannot measure its own performance is rarely a network that can hand an advertiser a defensible as-run file on request. The checklist below is ordered the way an audit actually unfolds — contract, clock, outage, panel, privacy, player — so you can send it to an integrator as-is.
Check 1 — Which record does the contract name as the source of truth?
Before comparing hardware, write down which log the advertiser or auditor will accept as the as-run record. If the contract is silent, you absorb the dispute risk, and no amount of good hardware will settle the argument later.
Distinguish the three candidates by evidentiary weight, because they answer different questions:
- CMS publish log — proves intent and scheduling. It shows what you sent and when you claimed it would play. It does not prove the screen rendered it.
- Player as-run log — proves delivery on that device, with a device-local timestamp, content ID and completion status. This is the record most advertiser contracts are actually describing.
- Panel acknowledgement or brightness telemetry — proves the presentation state: the screen was powered, awake and at the contracted brightness when the player says the spot ran.
Then test the plumbing, not the marketing. Request one anonymised sample as-run export (a single device, one week) with field definitions, and check whether it carries a stable device ID, content ID, start and end timestamps with a declared timezone, completion status and error codes — retained on-device as well as in the cloud, and exportable in a machine-readable format. A cloud-only log with no local retention is a single point of failure; if your CMS is still a mix of USB updates and IP distribution, resolve that architecture question first — the cloud CMS vs USB migration guide covers the transition and what it changes about where logs live.
Check 2 — Clock integrity: who owns the time source?
A timestamp from an unsynced clock is not evidence. Ask the supplier directly: does the player sync via NTP, from which named source, at what interval, and what does it write when that source is unreachable? An unstated drift figure is an unmanaged audit risk — the acceptable drift is your contractual decision, so require the supplier to declare the maximum drift they guarantee between syncs rather than accepting "it syncs automatically".
Cover timezone and daylight saving explicitly. Log in UTC with a declared offset, or your US and EU sites will disagree about what "played at 14:00" means across a DST boundary, and the disagreement will surface in the one month of the year you cannot afford it. Finally, ask what happens after a long power-off: does the player recover time before writing the next as-run entry, or does it backfill entries stamped with an unreliable clock? The second behaviour is common and it quietly poisons a week of records.
Check 3 — Offline behaviour, store-and-forward depth and replay accounting
This is the section competitors skip. Walk the supplier through two scenarios and demand the exact log output.
Scenario A: power cycle mid-loop
The store loses power during a scheduled slot; the player reboots and resumes the loop. What must the log contain? A clean termination or interruption entry for the slot in progress, a boot event with the new session ID, and — critically — a stated de-duplication rule so that the replayed segment is not counted and billed twice as two plays. Ask for that rule in writing. Manual reconciliation after every store-level power event does not scale past a few hundred screens.
Scenario B: a six-hour WAN outage
The network drops but the store stays open. Confirm three things: how many hours or days of scheduled content the player caches locally; how many hours or days of as-run records it buffers; and whether offline entries are written with a marker that distinguishes them from network-confirmed plays. Then ask how those local entries are reconciled with cloud records on reconnect — which record wins when the two conflict, and whether the merge is logged as an event or silently overwrites. A player that plays through an outage but loses the evidence of doing so has delivered nothing you can bill.
Check 4 — What the panel actually delivered
A datasheet brightness figure is a component figure, not a delivered figure at the glass. Buyers routinely specify screens by a headline number — "2500 nits" is a common search — but that number says nothing about whether the panel delivers it evenly across its face, and an impression claim rests on the delivered condition.
Request, in writing and with test conditions:
- Measured luminance at the glass, with the measurement method and conditions stated.
- Nine-point luminance uniformity across the panel face. A screen 15% darker in one quadrant looks like a different media placement on a video wall or menu board.
- White-point deviation, so colours do not drift between units in the same flight.
- Auto-dimming behaviour: is ambient-light dimming enabled by default, and is the brightness state written to the log? If brightness steps down at night and that state is unlogged, the delivered condition differs from the contracted one and you cannot show it.
- Ownership of the factory test report — does it ship with the unit or only on request?
For fixed retail installations, the mounting and enclosure also change the delivered result; the wall-mounted advertising display guide is a reasonable cross-check on how enclosure and placement affect what reaches the viewer.
Check 5 — Audience cameras: what you may log and what you must not
Keep two data classes architecturally separate: delivery logs (device and content metadata) and any sensor-derived audience data (counts, dwell, demographics). Mixing them is how a routine as-run request turns into a privacy review.
For any camera or sensor layer, specify on-device processing that outputs aggregate counts only — no images retained, no face templates, no biometric identification. Pin down notice (what signage or in-store notice is displayed), retention period for aggregate counts, which party is the controller under the applicable regime, and whether audience data leaves the store and to which jurisdiction. Confirm that legal position with counsel rather than relying on a vendor summary, and verify the layer can be disabled without breaking proof of play. Market-entry documentation is a separate question from auditability: model-level CE, FCC or RoHS declarations belong in your compliance file, and the certification checklist for LCD signage displays explains how to read them without over-claiming scope.
Check 6 — The player question your integrator will raise
Logging fidelity is a software and SoC question, not a panel question. Confirm the player can run the CMS agent version you need, export logs locally, hold a real-time clock across power loss, and expose its OS and agent version in the log header — so any audit can reconstruct the software state at the time of play. Ask whether log schemas change between firmware versions, because a schema change mid-contract can invalidate historical comparability across a fleet.
If an entry-tier SoC is proposed, check its documented decode and workload limits before committing; under-specified players are where as-run gaps appear. The RK3566 player limits comparison is the practical reference for where that tier stops being sufficient for a logging and decode workload.
The pre-purchase verification checklist to send your integrator
- One sample as-run log export — one device, one week — with field definitions.
- Declared time source, sync interval and guaranteed maximum drift.
- DST and timezone handling, and whether logs are UTC-normalised.
- Offline buffer depth in hours, stated separately for content and as-run records.
- Reboot and replay de-duplication rule, in writing.
- Reconnect reconciliation rule and conflict-resolution winner.
- Measured luminance at glass, nine-point uniformity figure and white-point deviation, with test conditions.
- Auto-dimming default state and whether brightness state is logged.
- Log retention period on-device versus in-cloud, and the export format.
- Firmware and log-schema change policy, plus the firmware update path.
Where proof of play ends and the buying decision begins
Be honest about the trade-off. Stronger logging carries a modest per-endpoint cost; a single disputed DOOH campaign can cost more than that delta across a store rollout — but you are buying risk reduction, not a guaranteed saving, so price it as insurance against a dispute you cannot currently answer. And restate the boundary: delivery evidence is defensible, while audience inference is an estimate with notice, retention and no-biometric-identification obligations attached to it. Vendors that blur the two are selling you a measurement claim your contract cannot support.
Then make the inquiry specific. Ask for a hardware specification that answers every item on the checklist above — clock behaviour, offline buffer depth, log schema, panel uniformity — for the screen count and store format you are actually deploying. Request the proof-of-play hardware specification sheet and ask for a quote; start with the player-and-panel specification comparison so your integrator is quoting against measurable requirements rather than a datasheet headline.
FAQ
Does digital signage proof of play prove anyone actually saw the ad?
No. It is a delivery record — the ad ran, when, and on which device. Whether it influenced a purchase requires a separate measurement layer with its own limitations and privacy obligations.
What timestamp source should a signage player use?
A network time source (NTP) with a declared source, declared sync interval and a supplier-declared maximum drift, with UTC-normalised logging and a declared offset. Ask what the player writes when the time source is unreachable.
What happens to proof-of-play records when the store loses internet?
It depends entirely on the player. A compliant one continues scheduled playback, writes local as-run entries with an offline marker, buffers them for a declared number of hours, and merges them with cloud records on reconnect under a written rule.
Can a CMS publish log alone satisfy an advertiser audit?
Rarely. A publish log proves what was sent, not what rendered. If your contract names only the CMS dashboard as evidence, you are accepting the risk of every device-side failure the dashboard cannot see.
Do proof-of-play cameras need consent, and how long may data be kept?
That depends on jurisdiction, the data captured and how it is processed. Aggregate, on-device counting with no image retention and no biometric identification reduces exposure, but notice wording, retention periods and controller status must be confirmed with counsel before deployment.
Wintouch Engineering Team
Our commercial-display engineers and product team review specifications against current factory records, deployment requirements and published standards. Learn more about our engineering capability and manufacturing operation.
Continue researching
Product facts and engineering claims follow our data and editorial policy.
Have a deployment question?
Share your environment, screen size and integration requirements with our engineering team.
