A multimeter will tell you whether voltage is present on a PoE port. It will not tell you whether the port can deliver the power a device needs. Understanding that limitation is the difference between a diagnosis and a guess.
First, the safety basics. PoE voltage is nominally 48 to 57 V DC, which is not dangerous to touch but will damage equipment if you short the wrong pins. Never probe a port with a device connected and the link active. Work with the port disabled or the cable unplugged, and use probes with fine tips so you do not bridge adjacent contacts.
Know which pairs carry power. 802.3af and 802.3at use either the data pairs, pins 1-2 and 3-6, called alternative A, or the spare pairs, pins 4-5 and 7-8, called alternative B. 802.3bt Type 3 and Type 4 use all four pairs for power. If you measure pins 1-2 and see nothing, check 4-5 and 7-8 before concluding the port is dead.
The measurement procedure for an idle port is simple: set the meter to DC volts, put the black probe on one conductor of a pair, the red probe on the other conductor of the same pair, and read. You should see somewhere between 44 and 57 V on a port that is offering power. Repeat across each pair according to the standard you expect.
Now the important part: **an open-circuit reading does not prove the port works.** PoE is a negotiated system. A correct PSE waits for a valid detection signature, roughly 25 kohm, before applying full voltage, and it may only apply a low test voltage until then. A port that shows 48 V open-circuit may still refuse to power the actual device, because the device fails detection, draws more than its class allows, or the PSE classifies it incorrectly.
To verify real delivery you need a load. Three practical options, in order of usefulness.
A dedicated PoE tester. These negotiate with the PSE, request a class, apply a load and display delivered voltage, current and wattage. This is the only handheld tool that answers the real question, and it costs less than one site visit.
An inline tester placed between switch and device. It shows live voltage and current while the actual device runs. Best for diagnosing a camera that boots then drops: you see the current spike at IR switch-on and can tell whether the port is current-limiting.
A resistive load with a meter. Workable but crude: you must know the resistance and do the arithmetic, and the load will not perform PoE detection, so many ports will never apply power to it.
Interpreting results. If voltage is present but the device does not boot, check in this order: device class against port class, cable pairs and continuity, then the port power budget. If voltage collapses under load, the PSE is likely current-limiting, which points at either a device drawing more than its class or a port already near the budget ceiling. If voltage is missing entirely on a port that had worked, check the port PoE status in the management interface before touching the cable.
Three mistakes account for most false conclusions. Measuring with the device connected and expecting a stable reading while the camera cycles its IR. Concluding "the port is dead" from a single pair reading on a bt port. And measuring at the camera end of a 90-metre run without accounting for voltage drop, then blaming the switch for a cable problem.
A last field habit worth adopting: log the measured voltage and current at commissioning for each PoE port, and write it on the port label. When a camera fails a year later, that baseline tells you instantly whether the fault is the device, the cable or the switch.