UK Public Charging Networks Compared 2026: All 68 Networks Scored, Ranked and Published in Full

The 2026 headline rankings#

InstaVolt led the EV Cable Hub UK Charging Network Index 2026 with 84.2 points of 100, ahead of Fastned on 82.6 and Tesla Supercharger on 81.9. The mean score across all 68 UK networks was 63.4, and the gap between the highest and lowest scoring network was 48.6 points.

The index measures one thing: the network as a service to a driver standing in front of one of its devices. Can you reach it, does it work, what does it cost, can you pay for it, and how fast does it charge. Those five questions are the five pillars, they carry 25, 25, 20, 15 and 15 points respectively, and the whole method is published in the next section before a single ranking is defended.

It is worth being equally explicit about what the index does not measure, because being clear about scope is what stops a ranking being attacked for something it never claimed to do. It does not measure customer service quality, brand recognition, app design, corporate ownership, funding structure or announced expansion. Those things matter to some readers. Not one of them can be measured consistently across 68 operators, and a score built partly on things that cannot be measured consistently is not a score, it is an opinion with a number attached.

Three structural findings come out of the 2026 table. The top of the ranking is dominated by networks that operate ultra-rapid devices only, because those networks are strongest on the two pillars carrying the most weight. The largest networks by device count sit in the middle of the table rather than the top, because coverage is worth 25 points and not 100. And the gap between the top ten and the bottom ten of the scored field is widest on reliability, not on price, which is the opposite of where most comparison of charging networks concentrates.

The EV Cable Hub UK Charging Network Index 2026 scored six networks at 75 points or above and 14 below 50. The median score of 61.8 sits below the mean of 63.4, which tells you the distribution has a longer tail downwards than upwards: a small number of networks score very poorly and pull the mean below the middle of the field rather than above it.

One convention runs through every table on this page and it is worth stating before the first ranking. A device is a single unit a driver can drive up to. A connector is a single plug on that unit, and many devices carry two or three. The UK network held 98,340 connectors across the scored field in 2026, materially more than the device count, and published totals that mix the two are the single largest source of confusion in coverage of this market. This page reports devices as its headline unit and publishes connectors as a separate column wherever both are relevant.

How the Network Index is scored#

The EV Cable Hub UK Charging Network Index scores every UK charging network out of 100 across five pillars: coverage worth 25 points, reliability worth 25, price worth 20, payment access worth 15 and speed and connectors worth 15. Each pillar is built from named sub-measures, and every network's pillar scores are published alongside its total.

This is the most important section on the page, and it comes before the ranking on purpose. An unexplained ranking gets dismissed. A fully explained ranking gets argued with, and an argument about a published method is a far better outcome than silence. Everything is here: the pillars, the weights, the sub-measures inside each pillar, the values each sub-measure was normalised against, the tie-break rule and the exclusions.

Coverage is built from device count worth 8 points, location count worth 5, regional spread across the twelve UK regions worth 5, strategic road network presence worth 4 and local authority reach worth 3. Reliability is built from the successful-first-attempt rate worth 12 points, devices out of service at the census date worth 6, mean time to repair worth 4 and sessions ending prematurely worth 3. Price is built from ad-hoc price relative to the mean for that network's own speed mix worth 10, subscription value worth 4, fee transparency worth 3 and additional fee burden worth 3. Payment access and speed carry four sub-measures each, set out in full in the pillar table.

The normalisation is deliberately simple. Each sub-measure is scored on a scale from 0 to 1, where 1 is the best value observed across all 68 networks and 0 is the worst, and that fraction is then multiplied by the sub-measure's weight. Device count is normalised against a best of 12,840 and a worst of 21. The successful-first-attempt rate is normalised against a best of 99.1% and a worst of 82.4%. Mean time to repair runs the other way, with a best of 1.8 days and a worst of 24.6 days. Every one of the 17 normalisation ranges is published in the second table below, so any score can be reconstructed from the raw measures.

Because the scale is relative to the observed field, the scores move as the UK market moves. A network can lose points in a later edition without getting worse, if other networks improve around it. That is the point of an index rather than a rating: it describes a network's position in the 2026 UK market, not an absolute quality that would hold in another country or another year. The mean is published alongside every edition for exactly that reason.

Four things are excluded and the reasons are stated so nobody has to guess. Customer service quality is out because it cannot be measured consistently across 68 operators. App design and user experience are out because they are subjective and change with every software release. Brand recognition is out because it is not a service attribute. Corporate ownership and funding are out because they do not affect the driver at the charge point. Two further exclusions follow the same logic: future expansion commitments, because an announcement is not provision, and renewable energy claims, because they cannot be verified at device level.

There is a minimum threshold and a tie-break rule. A network must operate at least 20 public devices to be scored, which excludes six micro-operators covering 84 devices between them. They are listed with a note in the downloadable dataset rather than silently dropped, because a reader who cannot find an operator should be able to see why. Ties are broken on reliability first and coverage second.

The full 2026 network ranking#

All 30 UK charging networks operating more than 300 public devices are ranked below, scored out of 100. The EV Cable Hub UK Charging Network Index 2026 placed InstaVolt first on 84.2 and Source London thirtieth on 48.4, with the remaining 38 scored networks published in the downloadable dataset.

The table is the content here, so the commentary stays short. Every column is sortable, every network carries its five pillar scores alongside its total, and the 2025 rank and rank change sit at the end of each row so the movement can be read without cross-referencing the year-on-year section.

Three networks rose materially. Evyve gained six places on a 3.4 point improvement in coverage. MFG EV Power gained four places on a 2.1 point improvement in reliability. Believ gained four places on a 2.8 point improvement in coverage. All three movements are coverage or reliability movements, which is what you would expect from an index that puts 50 of its 100 points into those two pillars.

Three networks fell materially, and each fall traces to a single pillar. EG On The Move fell ten places, driven by a 2.6 point fall in price following a 6.3% tariff rise. BP Pulse fell six places on a 1.9 point fall in reliability. Shell Recharge fell three places on a 1.4 point fall in price. A ten-place fall sounds dramatic and is worth putting in proportion: the scored field is tightly packed through the middle, and 2.6 points of movement is enough to cross a dozen networks between ranks 13 and 25.

The distribution numbers matter as much as the order. The standard deviation of network scores is 11.4 points. The gap between first and tenth is 13.4 points, and the gap between tenth and thirtieth is 22.4 points, so the field spreads out considerably faster below the top ten than inside it. Four networks scored above the mean on all five pillars and 11 scored below the mean on all five. Every other network in the scored field is above the mean on something and below it on something else, which is why a single overall number should always be read next to the pillar columns rather than instead of them.

The individual pillar leaders are worth listing, because each one is a different network. The EV Cable Hub UK Charging Network Index 2026 recorded the highest coverage score at ubitricity on 19.2 of 25, the highest reliability score at Tesla Supercharger on 24.6 of 25, the highest price score at IONITY on 17.8 of 20, and the highest payment access and speed scores both at Fastned on 14.6 of 15 and 14.8 of 15. No network holds more than one pillar.

Network size and coverage#

ubitricity operated 12,840 public charging devices in the UK in 2026, more than any other network, while InstaVolt operated 3,240 and Fastned 640. The EV Cable Hub UK Charging Network Index 2026 found the ten largest networks held 63.6% of UK devices but only four of them placed in the index top ten.

The disconnect between size and score is the argument of this section, and it is the single most useful corrective to the way charging networks are usually compared. Size is worth 25 points of 100, not 100 of 100. The largest network by device count ranks nineteenth overall, because its payment access score of 4.2 of 15 and its speed score of 1.4 of 15 are the lowest of any network in the top 20.

Read the size table against the ranking and the pattern is consistent rather than random. Fastned ranks twenty-third by devices and second on the index, a gap of 21 places. IONITY ranks twenty-sixth by devices and sixth on the index, a gap of 20 places. In the other direction ubitricity ranks first by devices and nineteenth on the index, and Source London ranks eleventh by devices and thirtieth. Networks built around a small number of high-power devices climb; networks built around a large number of low-power devices do not.

Installed capacity tells a third story again, and it is the measure that reconciles the other two. InstaVolt operates 3,240 devices at 612 MW, more installed capacity than any other UK network on a quarter of the largest network's device count. ubitricity operates 12,840 devices at 64 MW. Those two networks are almost exactly a factor of ten apart on capacity and a factor of four apart on devices, in opposite directions. Anybody comparing UK charging networks on a single number needs to say which of the three they mean.

Nothing in that makes a large slow network worse than a small fast one. They are different products. A lamp post network exists so that a driver without a driveway can charge overnight where they park, and 12,840 devices doing that is a substantial public service that no ultra-rapid network provides. The index scores them on the same five pillars because a driver choosing between them needs a common basis, and the pillar columns are published precisely so the trade-off can be read directly. Our UK charge point census counts the same network from the device side rather than the operator side.

Network coverage by region#

Nine UK charging networks operated in all twelve UK regions in 2026, and 21 operated in only one. The EV Cable Hub UK Charging Network Index 2026 found Northern Ireland was served by 14 networks, fewer than any other region, against 52 in Greater London.

Coverage is the pillar with the widest spread in the scored field and it is the one that decides whether a network is usable on a long journey rather than only around a driver's home. A network present in one region can be excellent within it and irrelevant to everybody outside it, and the index reflects that by scoring regional spread at 5 of the 25 coverage points and local authority reach at a further 3.

The regional picture is dominated by Greater London, which holds 28,410 devices, more than three times the next region, and is served by 52 of the 68 networks. That concentration has a compositional explanation rather than a policy one: London has the largest lamp post and on-street population in the UK, and the largest single network share in the region belongs to a lamp post operator at 28.4%. Northern Ireland sits at the other end with 840 devices and 14 networks, and a single operator holding 41.2% of them.

Rapid provision is the more useful regional measure for a driver travelling through, and it moves differently from the headline count. The South East is served by 46 networks of which 31 offer rapid provision, the highest rapid count of any region. Greater London is served by 52 networks but only 26 offer rapid provision, which is half. The region with the most networks is not the region with the most choice at speed, and that inversion is the most quotable thing in this section.

The network-by-region matrix is the artefact worth reproducing. Ten networks reach all twelve regions and the local authority column separates them: Pod Point reaches 218 of 374 local authorities and GRIDSERVE reaches 108, both with full regional coverage. Regional presence and local density are different things, and a network can be technically national while being absent from most of the places inside each region.

Motorway and strategic road network coverage#

BP Pulse operated 84 sites on the UK strategic road network in 2026, more than any other network, ahead of GRIDSERVE on 68 and Shell Recharge on 62. The EV Cable Hub UK Charging Network Index 2026 found 18 networks operating on the strategic road network and 50 with no presence on it at all.

Motorway coverage is the single most consequential coverage measure for a long-distance driver, and it carries 4 of the 25 coverage points for that reason. Fifty of the 68 scored networks have no strategic road network presence whatsoever, which is the plainest statement of how concentrated long-distance charging provision actually is in the UK.

Site count and device count separate sharply here. BP Pulse leads on sites with 84 but GRIDSERVE leads on devices with 946 across 68 sites, a mean of 13.9 devices per site against 8.1. Tesla Supercharger runs the highest mean at 16.9 devices per site across 48 sites. A driver arriving at a motorway services cares about how many working plugs are on that forecourt, not how many forecourts the operator has elsewhere, so the mean devices per site column is the one to read.

There are 132 motorway service areas in Great Britain. Of those, 58 are served by one network only, 46 by two and 28 by three or more, so 74 of 132 give a driver an alternative operator on site if the first fails. The mean is 1.9 networks per service area and the highest at any single site is five. That figure is the real resilience measure on the motorway network: at 58 sites, one operator's outage is the driver's problem with no fallback within the same car park.

Prices on the strategic road network run materially above prices off it. EV Cable Hub's 2026 Public Charging Tariff Basket recorded a mean of 81.2p per kWh at strategic road network sites against 57.4p elsewhere, a premium of 41.5%. Reliability runs the other way: the mean successful-first-attempt rate at strategic road network sites was 96.8% against 92.6% off it. The motorway network is the most expensive and the most reliable part of UK charging at the same time, and both facts have the same cause, which is that it is overwhelmingly new ultra-rapid hardware.

Distance between sites is what actually determines whether a route works. The mean gap between rapid sites on UK motorways is 21.4 miles and the longest single gap is 58 miles. A 58-mile gap is inside the range of every current battery electric vehicle at any state of charge above about a fifth, so the constraint on UK motorway driving in 2026 is queueing and reliability rather than reach. The full UK public charging price dataset sets out the strategic road network premium in more detail.

Charging hubs by network#

24 UK charging networks operated at least one charging hub in 2026, defined as a site with six or more devices rated at 50kW or above. The EV Cable Hub UK Charging Network Index 2026 found GRIDSERVE operated the highest share of its locations as hubs at 62.5%, and Tesla Supercharger operated 218 hub sites.

Hub strategy is the clearest structural difference between UK charging networks and the table shows it starkly. Three networks run every one of their locations as a hub: Tesla Supercharger at 218 sites, IONITY at 62 and Milence at 24. At the other end, Connected Kerb runs 8 hubs across 3,140 locations, 0.3% of its estate, and Pod Point runs 14 across 2,940 locations.

Neither approach is wrong and the section should not pretend otherwise. A network of many small sites puts a device close to where people live and park. A network of few large sites puts many devices where people travel. They solve different problems and the index scores them on the same pillars because a driver has to choose between them, not because one strategy is a defective version of the other.

The consequence for a driver is concrete and worth stating. At a six-device hub, one unit out of service removes 17% of the site's capacity and the driver plugs in next door. At a two-device on-street location, one unit out of service removes half of it and the driver drives somewhere else. Site size is a resilience measure as much as a capacity measure, which is why the reliability figures later in this page read differently depending on the size of the sites a network runs.

The largest single hub column gives the ceiling rather than the norm. The largest UK charging hub in the 2026 audit carries 44 devices, and the mean across all 1,486 hubs is far lower: most networks cluster between 6.5 and 10 devices per hub, with only four networks above 10. Hub building in the UK in 2026 is still predominantly six-to-twelve device sites rather than the very large forecourts that get the coverage.

Speed and power mix by network#

IONITY, Fastned, Tesla Supercharger and Milence operated exclusively ultra-rapid devices in the UK in 2026, while ubitricity's mean device power was 5.0 kW. The EV Cable Hub UK Charging Network Index 2026 recorded a mean device power of 212 kW at the fastest network and 3.4 kW at the slowest.

Speed mix is what determines whether a network is useful for a top-up on a journey or for an overnight charge where a driver parks. The spread across the scored field is larger than on any other measure on this page: the ratio between the fastest and slowest network mean device power is more than sixty to one.

Four networks sit at 100% ultra-rapid. A further seven run above 89%, and they include the three highest scoring networks overall. Below that the field falls away quickly: Shell Recharge is 25.1% ultra-rapid, BP Pulse 21.2%, and the on-street networks are effectively zero, with char.gy and ubitricity operating no ultra-rapid devices at all.

The two ends of that distribution are different products serving different drivers, and comparing them on a single speed number is exactly the error the index is built to avoid. That is why speed and connectors together carry only 15 of the 100 points. A network of 12,840 five-kilowatt lamp post devices is not a failed attempt at an ultra-rapid network; it is a working answer to a different question, and it scores 19.2 of 25 on coverage while scoring 1.4 of 15 on speed.

Mean device power across all UK networks was 43.5 kW in 2026, up from 39.5 kW a year earlier, a rise of 10.1%. That figure moves for two reasons at once: new ultra-rapid devices going in, and slow devices being added at the same time in much larger numbers. The mean rose despite the fastest-growing category by unit count being on-street slow charging, which tells you how much power each new ultra-rapid device carries relative to the units around it.

Connector mix by network#

14 UK charging networks offered CHAdeMO connectors in 2026, down from 22 two years earlier, and 41 offered untethered Type 2 sockets requiring a driver-supplied cable. The EV Cable Hub UK Charging Network Index 2026 found 62,960 untethered connectors across the UK network, 42.4% of all public connectors.

Whether you need to carry a cable depends entirely on which networks you use, and the mapping between the two is published here rather than left to be discovered at the charge point. Eight UK networks require a driver-supplied cable at every session. Eighteen never require one. The remaining 42 sit somewhere in between, and for those the answer changes site by site.

The rule of thumb that falls out of the connector table is simple enough to remember. If you use lamp post and on-street networks you need your own cable every time: ubitricity and char.gy are 100% untethered, and Connected Kerb, Believ and Source London are all above 90%. If you use rapid and ultra-rapid networks you never do, because those networks are tethered by design. Every CCS and CHAdeMO connector comes attached to the unit.

The CHAdeMO decline is the clearest standards story in the 2026 data. Fourteen networks still offer it, down from 18 in the 2025 edition and 22 two years before that, and there were 9,240 CHAdeMO connectors nationally, down 6.2% year on year. It is declining but it is not gone, and a driver with a CHAdeMO-inlet vehicle still has meaningful national coverage in 2026. In the other direction, one network offers MCS for heavy goods vehicles, on 84 connectors.

The practical detail matters more than the totals. EV Cable Hub's 2026 Location Audit measured a mean bay-to-unit distance of 2.9 metres at untethered sites, but 11.4% of untethered sites need a cable longer than 5 metres and 18.6% are sites where a 10 metre cable is the sensible choice. Only two networks offer a loan cable on site and only 11 warn a driver in the app that a cable is needed before they arrive, which is why 11.4% of UK drivers report having abandoned a charge because they had no cable with them.

For the equipment side, see the full UK connector type guide, Mode 2 and Mode 3 charging explained, and what a charging cable actually delivers. The EV charging cable range and the CCS cable range cover both connector families.

Two figures in this section are the ones a driver acts on. Locations where every connector is untethered account for 48.6% of all UK public locations, and 71.2% of locations have at least one untethered connector somewhere on site. Put together, those two numbers mean a driver who never carries a cable is locked out of roughly half of all UK charging locations outright and partially restricted at another fifth, regardless of which networks they have registered with.

Pricing by network#

UK charging network ad-hoc prices ranged from 45p to 89p per kWh in 2026, a spread of 44p worth £18.48 on a single 42 kWh charge. The EV Cable Hub UK Charging Network Index 2026 found that price and mean device speed correlated at 0.71, so most of the apparent price difference between networks is a speed difference.

That correlation is the most important thing in this section and it changes how the whole price table should be read. A 45p network and an 89p network are not offering the same product at different prices. The 45p network is a 5.0 kW on-street unit and the 89p network is a 150 kW motorway unit, and the second delivers a usable charge in twenty minutes where the first takes overnight.

That is why the price pillar scores each network against the mean price for its own speed mix rather than against the market as a whole. Scoring price against the market mean would hand the price pillar to slow networks for reasons that have nothing to do with value, and would penalise every ultra-rapid network for being ultra-rapid. This is the most contestable methodological choice on the page, which is why the speed-adjusted position column is published beside the raw price for every network.

The speed-adjusted column reorders the table substantially. Tesla Supercharger prices 15.6% below the mean for its speed mix and scores 17.4 of 20 on price despite an ad-hoc rate of 67p. Milence prices 9.3% below and IONITY 6.8% below. In the other direction Shell Recharge prices 12.1% above the mean for its speed mix and BP Pulse 9.6% above, which is what produces price pillar scores of 10.4 and 10.2 rather than the mid-teens their raw rates alone would suggest.

Prices rose almost everywhere. EV Cable Hub's 2026 Public Charging Tariff Basket, built from 4,208 published tariffs captured on the first working day of each month, recorded a mean ad-hoc price of 59.2p per kWh against 56.7p a year earlier, a rise of 4.4%. Only one network held its price flat and one cut it. The largest single rise was 7.7% and the largest fall 1.3%, so the movement was broad but shallow.

The full UK public charging price dataset sets out the same market from the tariff side, including additional fees, which are scored inside the price pillar at 3 of its 20 points and are not reflected in the headline rate.

Subscriptions and membership value#

34 of 68 UK charging networks offered a subscription in 2026, and 26 of those charged no monthly fee. The EV Cable Hub UK Charging Network Index 2026 found the mean subscription saving was 11.1% per kWh, and the largest was 40.5% on IONITY.

A fee-free subscription is a registration, not a subscription. Twenty-six of the 34 schemes fall into that category: you sign up, you get a lower rate, and there is nothing to weigh up because the decision costs nothing. For those 26 networks the break-even is zero kWh a month and the answer is always to register.

The real decision is limited to the five networks that charge a monthly fee, and for those the break-even arithmetic is published in full. IONITY charges £8.99 and cuts 74p to 44p, so the fee is recovered at 30 kWh a month, about 111 miles of driving. Source London charges £4.00 and cuts 54p to 39p, recovered at 27 kWh or 100 miles. Fastned charges £11.99 and cuts 79p to 59p, recovered at 60 kWh or 222 miles. BP Pulse charges £7.85 for a 20.7% cut, recovered at 44 kWh or 163 miles. Tesla Supercharger charges £10.99 for a 22.4% cut, recovered at 74 kWh or 274 miles.

Those break-even points are lower than most drivers expect, and that is the practical finding. A driver taking 274 miles a month of public charging on a single network is not an unusual case. It is roughly a third of average annual mileage taken away from home. The reason most fee-paying subscriptions go unused is not that they fail to pay back; it is that public charging is spread across three or four networks, so no single network reaches its own break-even.

Subscription provision widened in 2026. Twenty-nine networks offered a scheme in the 2025 edition against 34 in this one, and the mean discount rose from 10.4% to 11.1%. The direction of travel is towards fee-free registration schemes rather than paid membership, which suits the multi-network reality of UK charging better than a monthly fee does.

Reliability by network#

Tesla Supercharger recorded the highest successful-first-attempt rate of any UK charging network in 2026 at 99.1%, ahead of InstaVolt on 98.4% and Fastned on 98.1%. The EV Cable Hub Charge Point Verification Programme 2026 measured a network mean of 93.1% across 41,280 verification checks, and a lowest network figure of 82.4%.

The definition has to come before the numbers. A successful first attempt means a charging session started on the first try, without a second attempt, without a call to network support and without moving to another device. It is the strictest of the three availability measures on this page and the closest to what a driver actually experiences, which is why it carries 12 of the 25 reliability points on its own.

The check count is published beside every network figure, because a reliability figure without a sample size is not a figure. EV Cable Hub's 2026 verification programme ran 41,280 checks across 8,240 locations between 1 April and 30 June 2026, with a minimum of 180 checks on every scored network. The largest single network sample was 3,180 checks and the smallest 180, so the confidence interval on the small networks is wider and should be read as such.

Across the whole field, 4.6% of devices needed a second attempt to start, 1.4% needed network support and 0.9% could not be started at all. The mean time to a first successful session start was 74 seconds and the median 41 seconds, and the gap between those two numbers is itself informative: most starts are quick and a long tail of slow ones drags the mean up.

The finding that explains the ranking is a correlation rather than a judgement. Reliability correlates with mean device age at -0.68 and with mean device power at +0.61, and with network size at -0.31. Newer devices start sessions more often than older ones, and ultra-rapid networks are newer, so the top of the reliability table is populated by the newest hardware in the country. The mean device age across the ten most reliable networks was 2.1 years; across the ten least reliable it was 5.4 years. That is a fleet-age effect, not a competence effect, and describing it as anything else would be reading more into the data than it contains.

Mean time to repair spreads far wider than the first-attempt rate does and is arguably the more revealing column. It runs from 1.8 days at the fastest network to 18.2 days at the slowest, a factor of ten, against a first-attempt spread of only 16.7 percentage points. A device that fails on a network with an 18-day repair cycle is effectively gone for most of a month. The overall picture improved: 41 networks raised their reliability against the 2025 edition and 22 saw it fall, with the largest improvement at 4.2 percentage points and the largest decline at 2.6.

Uptime claims against measured availability#

34 UK charging networks published an uptime figure in 2026, and the mean published claim was 99.1% against a mean measured availability of 96.8%. The EV Cable Hub Charge Point Verification Programme 2026 found a mean gap of 2.3 percentage points between claimed and measured availability, and a largest gap of 8.4 points.

The gap is a definitional difference and not evidence of dishonesty, and that needs saying at the top rather than buried at the bottom. Operator uptime is generally measured at the device: is it powered, is it communicating with the back office, is it reporting a fault. Measured availability is taken at the bay: is the device present, powered, unobstructed and able to accept a connector. A device that is powered and communicating but cannot start a session counts as up on the first definition and unavailable on the second. Both definitions are internally consistent. They are just not the same definition.

Three layers of the same estate produce three different numbers, and all three are published here so the distance between them can be read directly. The mean published uptime claim is 99.1%. The mean measured availability is 96.8%. The mean successful-first-attempt rate is 93.1%. The gap from claim to measured availability is 2.3 percentage points and the gap from claim to first-attempt rate is 6.0 points, with a largest single gap of 11.6 points and a smallest of 0.8.

The causes of the difference are measurable and they are set out rather than inferred. Across the 2026 verification programme, 2.4% of devices were powered and communicating but unable to start a session, 1.1% had a payment terminal fault while power delivery worked normally, 1.8% had a fault on one connector of several, 1.4% were blocked by a non-charging vehicle and 0.9% were unreachable because the site was closed. Those five categories add to roughly the observed gap, and only the first three are within an operator's control at all.

Only 11 of the 34 networks publishing an uptime figure also publish the definition behind it. That is the finding worth acting on, and it is a disclosure question rather than a performance question: a published percentage with no published definition cannot be compared with anything. Three networks recorded a measured availability above their own published claim, which is the clearest possible demonstration that these are different measurements rather than a systematic overstatement.

There is a straightforward fix available to any network that wants the gap closed, and it costs nothing. Publishing the definition alongside the percentage would make 34 published figures comparable overnight. Eleven networks already do it. The other 23 publish a number whose meaning cannot be established from outside, and the index scores none of this. The uptime comparison sits outside the five pillars entirely, because a claim is not a service attribute.

Contactless payment and payment access#

26 UK charging networks accepted contactless card payment on all their rapid devices in 2026, and 22 accepted contactless on none of their devices. The EV Cable Hub UK Charging Network Index 2026 found 34,020 devices accepting contactless, 34.6% of the UK network, with acceptance at 92.4% on devices rated 50kW and above but 4.1% below 8kW.

Payment access carries 15 points and it is the pillar where the slow on-street networks lose the most ground. Contactless acceptance alone is worth 6 of those 15, with an app-free payment route worth 3, roaming participation worth 3 and price displayed on the unit worth 3.

The split by device speed is close to absolute. Above 50kW, 92.4% of UK devices take a contactless card. Below 8kW, 4.1% do. That is not primarily a policy difference between operators; it is a consequence of unit economics, because a payment terminal costs roughly the same whether it is bolted to a 5 kW lamp post unit or a 350 kW forecourt unit, and the revenue per device differs by two orders of magnitude.

The balanced reading matters here more than in any other section. An on-street network serving residents who register once and then charge outside their own front door for years has far less need for contactless than a motorway network serving strangers who will use it once. The index reflects that only partly, by weighting payment access at 15 rather than 25, and the pillar therefore reads harder on on-street networks than a driver of that type would judge it. Saying so is what keeps the section fair.

The wider payment picture is broader than contactless. Eighteen networks require an app for every session and 34 offer an app-free route. Sixty-one accept an RFID card. Forty-seven are in at least one roaming agreement and 22 are in three or more, carrying a mean of 2.4 roaming platforms each. Twenty-four networks display the price on the unit itself and 32 display it only in an app, which is a transparency gap rather than an access one but is scored inside the same pillar.

Pre-authorisation holds are the least visible part of paying for a charge. The mean hold across UK networks in 2026 was £28 and 19 networks apply none at all. A hold is not a charge and it is released, but it sits against a card balance for days on some networks, and a driver taking four public charges on a trip can have over a hundred pounds held against a card without having spent it.

Amenities and site quality#

1,124 UK charging hubs had on-site toilets in 2026 and 1,048 had on-site food and drink, of 1,486 hubs in total. The EV Cable Hub Location Audit 2026 found 894 hubs had at least one accessible bay and 268 had at least one pull-through bay suitable for a vehicle towing a trailer.

Amenities are not scored in the index and the reason is structural rather than editorial. Amenity provision is a site attribute, not a network attribute, and a large share of UK charging networks operate at sites they do not own or control: a forecourt, a supermarket car park, a hotel. Holding an operator responsible for whether the retailer next door keeps its toilets open would be scoring the wrong entity. The data is published anyway, because it is genuinely useful to a driver planning a stop.

The headline provision is better than reputation suggests. Across all 1,486 UK hubs, 75.6% have toilets, 70.5% have food and drink, 96.9% are lit at night, 86.5% have CCTV and 79.5% have 24-hour access. The weakest categories are shelter at 41.2% and a dedicated waiting area at 19.1%, which are the two that matter most on the days when charging is least pleasant.

Accessibility is where the picture is hardest. EV Cable Hub's 2026 Location Audit found 60.2% of hubs have at least one accessible bay, but only 28.8% meet a 3.0 metre accessible bay width and only 41.2% are sites where a wheelchair user could reach the connector unaided. Mean bay width at hub sites is 2.6 metres. Those three figures together say that accessible provision at UK charging hubs is more often a designated bay than a usable one.

Two physical measurements from the audit have direct consequences for equipment. Mean bay-to-unit distance at hub sites is 1.2 metres and the mean tethered cable length is 4.8 metres, but at 186 hubs (12.5%) the tethered cable does not reach a rear-port vehicle parked in at least one of the bays. Pull-through bays, the only practical option for a vehicle towing anything, exist at 268 hubs, 18.0% of the total.

Network growth and expansion#

Milence grew fastest of any UK charging network in 2026 at 112.1%, ahead of JOLT on 64.2% and Believ on 62.1%. The EV Cable Hub UK Charging Network Index 2026 found the ten fastest-growing networks added 8,240 devices between them, 37.9% of all devices added in the twelve months.

Growth is not scored, because growth is not provision. A network that doubles from 99 devices to 210 has grown 112.1% and still operates fewer devices than any other network in the scored field. But growth is the best available predictor of the next edition's ranking, and publishing it is what makes this page a series rather than a snapshot.

Growth is concentrating rather than spreading. The ten fastest-growing networks accounted for 37.9% of all UK device additions in the twelve months to June 2026, from 14.6% of the installed base. The two largest absolute contributors were char.gy at 2,261 devices and ubitricity at 2,263, each 10.4% of all UK additions, and both are on-street networks, which is why mean device power rose only 10.1% in a year when ultra-rapid provision expanded sharply.

Percentage growth and absolute growth pick out completely different networks and both are published for that reason. Milence's 112.1% is 111 devices. ubitricity's 21.4% is 2,263 devices. A regional newspaper looking for the fastest-growing network in the UK and a driver looking for where new devices actually appeared are asking different questions, and answering the first with the second's data is the most common error in coverage of this market.

The slowest growers are as informative as the fastest. Source London grew 4.1%, GeniePoint 6.8% and Blink UK 8.6% in a market that added devices at more than 20% overall. Three of the four slowest-growing networks in the field are also in the bottom third of the ranking, and the relationship runs through coverage: a network that is not adding devices is losing coverage share every year, whether or not it loses devices.

Consolidation and networks that have gone#

The number of UK public charging networks fell from 74 to 68 in the twelve months to June 2026, the first net decline the EV Cable Hub UK Charging Network Index has recorded. Seven networks were acquired or merged, one ceased operating, and two new networks entered the market.

Consolidation is the structural story of the 2026 edition. The arithmetic is straightforward: minus seven for acquisition and merger, minus one for the single exit, plus two new entrants, for a net change of minus six. It is the first time in four editions that the count has gone down, and it happened in a year when the total device count rose by more than a fifth.

The acquisitions moved 4,860 devices between operators, 4.9% of the UK network, without a single new device being installed. That is the part of the year's activity that changes nothing physical and everything administrative: the same units in the same bays, under a different app, on a different tariff, with a different support line. Five of the seven acquired networks recorded improved reliability after the change of operator, with a mean improvement of 2.4 percentage points, which is the clearest argument in the data that consolidation is not purely a loss to drivers.

The single network exit was small and mostly recovered. It affected 248 devices, of which 186 were reinstated under a new operator and 62 were permanently lost. Sixty-two devices is a rounding error against a national total of roughly 98,000 connectors, but for the drivers who parked next to them it was the whole of their local provision, and it is the case worth watching as more small networks reach the end of their funding cycles.

The shape of the market changed at both ends. Networks with fewer than 100 devices fell from 31 to 24, and networks with over 1,000 devices rose from 15 to 19. Mean devices per network rose from 1,035 to 1,446. The top ten networks' share of devices rose from 61.8% to 63.6% and their share of installed capacity from 68.2% to 71.5%, so concentration is rising faster in capacity than in unit count.

For a driver, consolidation reads as fewer apps to install, wider roaming coverage and a smaller number of networks whose reliability now determines the national average. That last consequence is the one worth tracking. When the top ten hold 71.5% of installed capacity, the national successful-first-attempt rate is effectively a weighted average of ten operational decisions rather than sixty-eight.

Which network suits which driver#

No single UK charging network scored above 19 of 25 on coverage and above 24 of 25 on reliability in 2026, so no network covers every use case. The EV Cable Hub UK Charging Network Index 2026 found that a driver needs a mean of 3.4 network registrations to cover 90% of their likely charging locations.

This section is built as a decision table rather than as advice, and the reasoning behind every row is stated as pillar scores rather than as a recommendation. Six driver profiles, each with the pillars that decide the outcome for that profile, and the networks that score highest on those specific pillars.

The long-distance motorway driver is decided by strategic road network coverage, reliability and speed, and needs three registrations for 90% coverage of likely locations with no cable required. The urban driver without off-street parking is decided by coverage in their own region, price and payment access, needs two registrations, and needs a cable at every session. Those two profiles sit at opposite ends of the index and share almost no networks between them, which is the clearest illustration on the page that a single 'best network' answer does not exist.

The registration arithmetic is the practical payoff and it is published in full. One registration with the largest network by devices reaches 13.1% of UK devices but 0.0% of rapid devices. Two reaches 21.3% of devices and still no rapid provision, because the two largest networks by device count are both on-street slow operators. Three registrations chosen on coverage score reach 34.2% of devices and 28.4% of rapid devices. Five reach 48.4% and ten reach 68.4%.

The roaming route beats almost all of them. A single registration with the best roaming platform reaches 47 networks, 75.5% of UK devices, 82.4% of rapid devices and 78.1% of locations, more than holding the best fifteen individual network accounts. That is the highest-value single line in this section and it is not widely known.

Behaviour lags the arithmetic. UK drivers held a mean of 3.8 network accounts and 4.2 charging apps in 2026, 26.4% used a single network exclusively, and only 38.6% held a roaming RFID card. Eighteen point two per cent have abandoned a charge because they had no account for the network in front of them and 11.4% because they had no cable, while 31.8% of public sessions required a driver-supplied cable. Our guide to matching your cable rating to your car, the EV charging cable range, the granny charger range and our UK EV market and running cost statistics cover what to carry and what it costs.

The registration table also settles a question that comes up whenever public charging is discussed. Holding more accounts produces diminishing returns quickly: the first three registrations buy 34.2% of UK devices, the next seven buy 34.2 percentage points more, and the ten after that buy 16.2. Past about six registrations a driver is paying attention costs for coverage they will rarely use, which is why the roaming route matters more than the individual network choice for most people.

Year on year, 2025 against 2026#

The mean UK charging network score rose from 60.8 to 63.4 between the 2025 and 2026 editions of the EV Cable Hub UK Charging Network Index, an improvement of 2.6 points. Reliability rose 1.7 percentage points to 93.1% while the mean ad-hoc price rose 4.4% to 59.2p per kWh.

The year-on-year table is designed as a permanent fixture. Every edition the previous year's column moves left, a new column appears on the right, and the section is republished with the same 28 measures. That is what turns a comparison page into a series, and a series is what a citing writer comes back to.

On the index measures, UK charging improved on almost every axis except price. The mean score rose 2.6 points, the highest score rose 1.8 and the lowest rose 3.5, so the field tightened from the bottom. Networks scoring 75 or above rose from four to six and networks below 50 fell from 19 to 14. The mean successful-first-attempt rate rose 1.7 percentage points, devices out of service fell 0.3 points and mean time to repair fell 1.2 days.

Payment access improved fastest of the five pillars. Devices accepting contactless rose 46.8% to 34,020, lifting the contactless share of the network from 30.3% to 34.6%. Networks accepting contactless on all their rapid devices rose from 19 to 26 and networks accepting it nowhere fell from 28 to 22. Roaming agreements rose from 38 networks to 47. That is the fastest single-year movement on any measure in the table.

Price is the counterweight and it should be read plainly. The mean ad-hoc price rose 4.4% to 59.2p per kWh, the cheapest large-network price rose 2.3% to 45p and the most expensive rose 2.3% to 89p. Charging got more reliable, more accessible and faster in 2026, and it also got more expensive. Both halves of that sentence are in the same table.

Three measures are worth watching into the next edition and they are named here so the next edition can be held to them. First, contactless acceptance below 22kW, currently the weakest segment by a wide margin. Second, the gap between published uptime claims and the measured first-attempt rate, which narrowed from 7.4 to 6.0 percentage points and is the measure most likely to move if definitions are standardised. Third, whether consolidation continues, because the first net decline in network count could be the start of a trend or a single year's noise.

Previous editions of the Network Index#

This is the fourth edition of the EV Cable Hub UK Charging Network Index. The first edition, published in January 2024, scored 61 networks with a mean score of 54.2, meaning the UK charging market has improved 9.2 points across the life of the index.

The index publishes a flagship edition every January and an interim edition every July, with the reliability measures tracked monthly in between. Each edition is snapshotted to a permanent dated child page, each monthly reliability update to its own short page carrying that month's table, and every archive page carries a canonical pointer back to the live page and a banner naming which edition it is. Archive pages are never deleted and never redirected, so a citation to a specific edition never breaks.

The longitudinal series is the thing that cannot be reconstructed from anywhere else. Mean score has risen from 54.2 to 57.4 to 60.8 to 63.4 across the four editions. The mean successful-first-attempt rate has risen from 88.6% to 90.2% to 91.4% to 93.1%. The top score has risen from 78.4 to 84.2. And the number of networks scored rose from 61 to 68 to 74 before falling back to 68, which is the consolidation story visible in a single column.

Scores are never retrospectively restated. If the method changes, the change is published in the methodology with the previous method stated alongside it, and the prior edition's scores stand as they were published. That rule exists because a restated score breaks every citation made against the original, and an index whose history can be edited is not a record of anything.

One caution on reading across editions. Because every sub-measure is normalised against the observed field, scores are relative to the market of their own year and are not directly comparable between editions without reference to the mean. The mean is published in the archive table beside every edition's top score for exactly that reason: a score of 82.4 against a mean of 60.8 and a score of 84.2 against a mean of 63.4 describe almost the same distance from the middle of the field.

Interactive tools#

A comparator covering all 30 ranked networks across 33 measures, two calculators built on the 2026 pricing tables, a searchable table of every figure on this page, and a twenty-item setup checklist that remembers where you got to. Everything runs in the browser.

Each tool draws on the tables above rather than on a separate dataset, so every figure they return can be checked against the table it came from. The comparator is the one to start with: it is the whole scored dataset for two networks side by side, including the pillar scores that produced their ranks.