Drivers judge a charging network by whether it works when they arrive, not by the power figure on the sign. The gap between those two things comes down to several unglamorous factors.
Uptime is measured in ways that flatter operators
An operator typically counts a charger as available if it reports itself online. A unit that answers the network but fails to deliver a session still appears healthy in that count.
Drivers measure something different: whether a charge was successfully completed on the first attempt at the first stall tried. Those two figures can diverge sharply.
The distinction matters because investment follows the metric being reported, and a network optimising for reported uptime may not be fixing what actually frustrates users.
Most failures are not electrical
Payment and authentication account for a large share of failed sessions. A card reader that cannot reach its payment processor will refuse a session even though the power electronics are perfect.
Cellular connectivity is a common weak point, particularly at sites chosen for road access rather than signal strength. A charger that cannot phone home usually will not start.
Cable and connector damage is the other frequent cause, since these are heavy items handled outdoors by the public in all weather and are exposed to being driven over.
Grid connection sets the real ceiling
A site's total power is fixed by its connection to the local network, and that connection is expensive and slow to upgrade. The rating of individual posts can exceed it comfortably.
When several cars charge at once, available power is shared. A stall advertised at a high figure may deliver a fraction of it simply because the neighbouring bays are occupied.
Some sites use battery storage to buffer this, drawing steadily from the grid and discharging quickly into cars, which raises peak capability without a larger connection.
Maintenance economics decide how fast faults are fixed
A charger in a remote location generates little revenue, so sending an engineer to it is disproportionately expensive relative to what it earns.
Networks with dense site coverage can service several units per visit, which shortens repair times. Thinly spread networks face the opposite arithmetic and repairs stretch out.
This is why reliability tends to correlate with network density rather than with the age or specification of the hardware installed.
Redundancy matters more than peak power
A site with several stalls is far more useful than a site with one very powerful one, because any single unit can be out of service without stranding the driver.
Queueing behaves the same way. Doubling the number of bays reduces waiting disproportionately, since arrivals are irregular rather than evenly spaced.
For journey planning, the number of working stalls at a location is a better predictor of a smooth stop than the headline kilowatt figure on the sign.