Why DOOH Screens Go Dark — Connectivity, Telemetry and Triage

10 min read · Sep 19, 2026

A dark digital display panel in a public space at night — Photo: Unsplash

A dark screen is never dramatic. It is almost never the software, and it is almost never the thing anyone budgeted for. It is a SIM that ran out of data on the 28th, a disk that filled with orphaned media, a cleaner who unplugged a socket to run a vacuum, or a panel that came back from a power cut and waited politely for someone to press a button. The work of running a screen network is not creative — it is knowing which of those happened without driving to the site to find out.

The failure modes, in the order you will actually meet them

Power. By a wide margin the most common cause, and the one most often mistaken for a software fault. A panel that does not auto-resume after an outage looks identical to a crashed player.
Connectivity. The link is up but useless — a captive portal has logged the device out, a data cap has throttled it to nothing, or a DHCP lease expired and nobody renewed it.
Storage. Media accumulates, temporary files never get cleared, and the player runs out of room to download the next playlist. It keeps showing the old content, so nothing looks wrong until the campaign that was supposed to start never appears.
Heat. A panel in a west-facing window in summer will throttle and eventually shut down to protect itself. This is seasonal and gets blamed on everything else.
The cable. An HDMI connector works loose behind a wall-mounted screen. There is no log entry for this.

The screen was fine. The screen is nearly always fine. It is the six things attached to the screen that are not.

Connectivity: what to put behind a screen

Wired ethernet is the right answer wherever you can get it, and the reason is not bandwidth — signage needs very little — it is that a cable does not silently deauthenticate. Where a cable is impossible, the realistic options are a dedicated 4G router or a modem built into the player, and the choice matters more than it looks.

A dedicated 4G router with its own SIM keeps the screen off the venue's guest WiFi, which is the single biggest source of mystery outages. Guest networks reset, rotate passwords and expire sessions on schedules nobody tells you about.
A built-in modem is tidier and one less thing to power, but you cannot swap it or move it to another device when the player fails.
Buy a data plan with a hard floor, not a cap that throttles. A throttled connection is worse than no connection, because the device still reports itself online.
If the site has WiFi you must use, put the screen on a separate SSID or VLAN if the venue will allow it, and get the credentials in writing before install day.

Whatever you choose, bandwidth is rarely the constraint. A player that caches content locally needs the network only to fetch updates and report back. The requirement is not speed, it is reliability — a link that is up 99% of the time beats one that is fast and up 90% of the time.

Telemetry: what is actually worth monitoring

Most monitoring dashboards report whether a device is reachable, which tells you almost nothing. A player can be perfectly reachable and showing a black screen, or reachable and playing content from three weeks ago. Reachability is the weakest possible signal and it is the one most platforms lead with.

Heartbeat with a timestamp — not just up or down, but when it last checked in, so you can tell a brief blip from a device that died on Tuesday.
Last successful playback, per item. This is the signal that catches the silent failure where a device is online but the content never rotated.
Free storage. Trend it rather than alerting at a threshold; a disk filling steadily tells you a month in advance.
Current playlist version against the assigned one. A mismatch means the device did not take the update and nobody knows.
Panel power state and, where the hardware exposes it, temperature.

The test of a monitoring system is not whether it can tell you a screen is down. It is whether it would have told you the campaign that launched on Monday never actually reached nine of your forty screens — while all forty reported healthy.

Reachable ≠ workingThe distinction most signage monitoring misses — and the one that costs an advertiser their campaign

Triage: a dark screen, and you are 200km away

Work outward from the cheapest check. Most faults resolve in the first two steps, and each step you can do remotely is a site visit you did not pay for.

Is the device reporting at all, and when did it last check in? A device that stopped reporting at 02:14 and a device reporting healthy right now are completely different problems.
If it is reporting — what does it think it is playing, and does that match what it should be playing? This separates a content-delivery fault from a display fault.
Try a remote reboot. This fixes a genuine majority of faults and costs nothing.
Check free storage and the last successful download. A full disk presents as stale content, not as an error.
If it is not reporting, establish whether the network or the device is out. If other devices at the same site are fine, it is the device; if they all went together, it is power or the link.
Only then send someone — and send them knowing whether to bring a SIM, a cable or a replacement player, rather than all three and a guess.

Design so an outage is not an incident

The most useful property a player can have is that losing the network does not stop the screen. A player holding its content locally keeps playing through the outage, and reconciles when the link returns — which converts most connectivity faults from an emergency into a line in a log. It also means a data cap hit on the 28th does not blank a storefront for three days.

The corollary matters for anyone selling advertising: if playback continues offline, the playback record has to continue too, buffering locally and uploading on reconnect. Otherwise every outage becomes a hole in the proof-of-play report, and the advertiser is entitled to ask why they should pay for hours you cannot evidence.

nextdooh players are offline-first — each one loops its local cache through an outage and reconciles afterwards — and record playback per second locally, uploading when the connection returns so the audit trail has no gap. Device telemetry covers heartbeat, storage, playlist version and last successful play, with alerts on the signals that matter rather than on reachability alone, and screens can be rebooted remotely from the dashboard.