Why EV Charger Uptime Is the Metric That Really Matters?

Public charging targets are usually written in chargers installed. Programs announce ports funded, transport ministries count connectors per mile of highway, and operators lead their launch releases with network size. Drivers keep a different score. For them, a site is only as good as the charger they actually plug into.

The gap between those two scores can be large. In early 2022, a team led by UC Berkeley’s David Rempel tested every open public fast-charging port with a CCS connector in the Greater San Francisco Bay Area, 657 connectors at 181 stations. Only 72.5% delivered a charge. The failures were mostly ordinary: blank screens, payment errors, sessions that would not start, network faults, broken connectors, and cables too short to reach the car. At the time, operators were reporting uptime of 95 to 98%.

Things have improved. J.D. Power’s 2026 U.S. EVX Public Charging Study, released in August, found that 12% of owners had recently arrived at a public charger and left without charging, down from 14% a year earlier and the lowest figure the study has recorded. That is real progress, but it still means roughly one driver in eight. For anyone funding, regulating or buying charging infrastructure, the useful question is not how many chargers were built. It is how many work when someone pulls in.

What uptime should actually measure

Ask a few operators what uptime means and you will get a few different answers.

The loosest version asks only whether the charger was connected to the back end. A charger can stay connected all day and still charge nobody: a connector jams, the card reader fails, or payment drops at the last step. On the operator’s dashboard, it shows green. At the curb it is broken.

A better test is whether a driver could have started and finished a paid session at that moment. That means measuring each connector rather than each charger, counting sessions that never got going, and treating a unit that keeps dropping off and reconnecting as a problem even if it always comes back. The Rempel study ended on the same point: the sector needs shared, precise definitions of uptime, downtime, and excluded time, checked by someone other than the operator.

Exclusions matter as much as the headline number. Most definitions leave out planned maintenance, and some leave out vandalism or grid outages. That is reasonable when maintenance is scheduled at quiet hours and shown in the driver app. But a driver who arrives during the window still cannot charge, and generous exclusions can make a weak network look compliant.

Uptime is now a funding condition

Regulators have started writing these definitions into law, which is why uptime has moved from an operations concern to a contract term.

The two rules differ in useful ways. NEVI measures each port, so one bad site cannot hide inside a large network. The UK rule measures the network but sets a higher bar. Both rely on the operator’s own status data, which brings back the definition problem above. A program that wants the uptime it pays for should specify connector-level availability and session success, and keep the right to verify them independently.

Why chargers go down

In J.D. Power’s 2025 study, a charger that was out of service or not working properly accounted for 60% of failed visits. The causes behind those failures are rarely dramatic.

Connectivity is a large share. Chargers need a cellular or wired link to the charge point management system (CPMS), and underground parking garages are often poor places for a cellular signal. A charger with a patchy connection can appear offline, or fail to authorize a session, several times a day with nothing physically wrong with it.

Firmware and configuration come next. A unit may need a reset after an update, or a setting changed during commissioning may quietly block one payment method. These faults are usually quick to fix once someone notices them.

Then there is wear and tear: cables driven over, connectors left on wet ground, the occasional smashed screen. Some downtime is not a fault at all. A car that finished charging an hour ago and is still plugged in makes that bay as useless as a broken unit.

In most of these cases, the operator learns about the problem from a driver complaint, by which point the charger may have been unavailable for most of the day.

How operators find faults before drivers do

Networks with strong uptime do not always run the newest hardware. What sets them apart is how quickly they detect and clear faults.

Detection comes first. If a charger has not reported in for a set period, the system should query it rather than wait for a failed session to reveal the problem, and alert the operations team as soon as a site drops off.

Remote recovery comes second. Many faults clear with a remote reset, a connector unlock or a configuration change, and a capable CPMS can run the known fixes automatically: reset, confirm the charger is back, and log the event.

Some faults still need a technician. Grading faults by severity means serious ones become work orders immediately, and the technician arrives knowing what to expect. A job should close only when the charger itself reports every connector as available, not when someone checks a box on a form.

Finally, there is the bay itself. Overstay rules, or an idle fee that starts once a session ends, keep connectors free for the next driver.

The metrics worth asking for

A single uptime percentage in a quarterly report says little. Program managers and asset owners should ask operators for these instead:

MetricWhat it tells you
Connector-level availabilityWhether each plug could be used, not just the charger as a whole
Session success rateThe share of attempted sessions that start and finish without errors
Time to detectHow long a fault exists before the operator knows about it
Time to recoverHow long from detection until the charger works again
Remote resolution rateThe share of faults fixed without a site visit
Repeat faultsChargers or sites that keep failing, often a sign of a connectivity or installation problem

If only one is added to a reporting template, make it time to detect. Where drivers are the ones finding faults, it is usually measured in hours, and bringing it down tends to lift every other number.

Questions for procurement

Much of the difference between a reliable network and an unreliable one sits in the operations layer: how quickly the operator hears about a fault and how much it can fix remotely. When a program, utility or property owner is selecting an operator or a CPMS, these questions are worth putting in the tender:

  • How does the system detect a charger that has stopped reporting, and how quickly does it alert staff?
  • Are faults graded by severity and turned into work orders automatically, and are known fixes tried remotely first?
  • Can field technicians record their visit on site and close a job only after the charger confirms it is working?
  • Does the platform support open standards such as OCPP for chargers and OCPI for roaming and data sharing, so the owner is not tied to one hardware brand?
  • Will the operator report the metrics above at connector level, and accept independent verification?

Risks run both ways. Strict uptime clauses with liquidated damages raise operating costs, and operators will price that into their bids. Contracts that define uptime loosely, on the other hand, pay for chargers that may not work.

Sources


About the author

Krati Tripathi is Content Marketing Manager at Zynetic Electric Vehicles Charging Private Limited, a company that develops charge point management software for EV charging operators. She writes about the operational side of EV charging: how faults are detected and fixed, how uptime is measured, and what operators, property owners, and public funders should ask for when they buy charging infrastructure.

Disclosure. The author is employed by Zynetic, which develops charge point management software of the kind discussed in this article. Zynetic did not pay for this article, and no products or vendors are recommended in it. AI tools (Anthropic’s Claude) were used to help draft and edit the text; the author reviewed and revised it, and the figures are drawn from the linked sources.


author avatar
Krati Tripathi Content Marketing Manager
Krati Tripathi is Content Marketing Manager at Zynetic Electric Vehicles Charging Private Limited, a company that develops charge point management software for EV charging operators. She writes about the operational side of EV charging: how faults are detected and fixed, how uptime is measured, and what operators, property owners and public funders should ask for when they buy charging infrastructure.
Verified by MonsterInsights