45% of UK Businesses Don't Know Who's Responsible When Broadband Fails

On 2 September 2026 a piece of UK connectivity research landed with a finding that most business owners will recognise instantly, and most connectivity contracts still fail to address. A survey of 401 full-time UK business connectivity decision-makers, commissioned by altnet provider LightSpeed Networks, found that 45% have already lived through a fault where responsibility between suppliers was disputed or unclear. Not a fault that was slow to fix, or expensive to fix, or badly communicated - a fault where the basic question of whose job is this could not be answered while the business sat offline. That is close to one in two organisations, and it describes a failure mode that has nothing to do with fibre, hardware or engineering skill.

The same research puts a second number underneath the first. 34% of buyers say that ownership of fault resolution across their connectivity supply chain is not clearly defined at all - meaning the ambiguity is not something that emerges in the heat of an incident, it is baked into the arrangement from the day the contract is signed. In those organisations, an outage does not trigger a process. It triggers a search for one. And it does so at precisely the moment when the business has the least capacity to conduct it: staff idle, card terminals dead, cloud applications unreachable, and a customer-facing clock running. Today that blind spot is getting more expensive rather than less, because 75% of the same buyers say cloud and AI adoption has made resilient, always-on connectivity more important to their operations than it used to be. The dependency has deepened; the accountability has not kept pace.

What the research actually found

The survey covers 401 full-time UK business connectivity decision-makers - the people who sign, renew and are held responsible for the circuits their organisations run on. It was commissioned by LightSpeed Networks, an alternative network operator, which is worth stating plainly: a company that owns and operates its own network has a commercial interest in research showing that buyers value owned-and-operated networks. That does not invalidate the numbers, but it does mean the sensible way to read them is as a description of buyer sentiment and experience rather than as a neutral audit of the industry. The experiential findings - what has actually happened to these organisations during outages - are the part that matters most, and they are the part hardest to spin.

The headline experiential finding is the 45%. Nearly half of respondents have been through an incident where responsibility between suppliers was disputed or unclear. That is a specific and unusual thing to measure. Most connectivity surveys count outage frequency, mean time to repair or satisfaction scores. This one counts something closer to the lived reality of a business incident: the interval during which nobody has yet accepted that the problem is theirs. In engineering terms it is dead time. Nothing is being diagnosed, no van is being dispatched, no configuration is being rolled back. The fault is static while the ownership question is resolved, and the business absorbs every minute of it.

The structural finding sits alongside it: 34% say ownership of fault resolution across their connectivity supply chain is not clearly defined, and that this sometimes forces the business itself to navigate multiple suppliers during an outage. Read that carefully, because it inverts the normal service relationship. The customer - the party with the least visibility of the network, no access to the fault management systems, no engineering relationship with the infrastructure owner and no contractual leverage over anyone except the company it pays - becomes the integrator of last resort. An office manager with a mobile phone ends up brokering a technical conversation between a reseller, an ISP and a wholesale network operator, none of whom report to them, while trying to run a business that has stopped working.

The third block of findings is about what buyers now say they want as a result. 61% said they prefer providers with an end-to-end, owned and operated network. 66% said knowing who owns the underlying infrastructure matters more to them than it did a year ago. And when asked what would make them leave, poor outage response was the single most-cited reason to switch, named by 56% of buyers - ahead of price at 51% and poor customer support at 48%. That ordering is the finding with the most commercial weight in the whole dataset. In a market that has spent a decade competing almost entirely on headline price and headline speed, buyers are now telling researchers that how a provider behaves on the worst day beats what it charges on all the other days.

How a disputed outage actually unfolds

The survey reports the pattern; it does not narrate it. The chronology below is a composite - a representative sequence assembled from the structure the research describes, not a record of one named incident - because the shape of these events is remarkably consistent across organisations. Times are shown from the moment the business first notices, which is almost never the moment the fault began.

What UK connectivity buyers say matters now

Set the survey’s figures side by side and a coherent buyer position emerges. It is not a demand for faster circuits or cheaper ones. It is a demand for a shorter chain and a named owner.

The ordering of the three switching triggers is the part worth dwelling on. Outage response at 56% outranks price at 51%. For most of the past decade the UK business connectivity market has behaved as though the reverse were true - comparison sites rank on monthly cost, tenders are scored heavily on price, and renewal conversations open with a discount. What buyers are now reporting is that a bad incident does more to end a supplier relationship than an expensive invoice does. That is a rational position rather than an emotional one. A price difference of a few pounds a month is a known, bounded, budgeted cost. A disputed outage is an unbounded one, and it lands on the days when the business can least afford it.

The 75% figure explains the shift in pressure. A business whose critical systems ran on a local server and a phone line could absorb a connectivity outage as an inconvenience - staff worked from local files, took orders on paper, and reconciled afterwards. A business running Microsoft 365, a cloud finance platform, a hosted phone system, cloud-based card processing and an increasing quantity of AI-assisted work has no offline mode at all. Every one of those dependencies converts a connectivity outage directly into a full operational stop. The circuit stopped being a utility and became the platform, and the accountability model around it never caught up.

Why the contract, not the cable, decides who is responsible

There is a clear answer to the question in the headline, and it is worth stating before anything else, because a large number of UK businesses do not know it. Ultimate responsibility for an outage sits with whichever company holds the customer’s service contract. That company must then engage whoever owns the underlying infrastructure. It does not matter that the retail provider did not lay the fibre, does not operate the exchange and cannot dispatch the engineer directly. The obligation follows the contract, and the contract is with the party that invoices you.

The single exception is a vertically integrated provider that owns the full stack itself - where the company selling you the service is also the company that owns and operates the network carrying it. In that case there is no hand-off, because there is no second party to hand off to. That structural fact is precisely what sits behind the 61% preference for end-to-end owned and operated networks and the 66% who now care more about who owns the infrastructure. Buyers are not expressing a technical preference about network architecture. They are expressing a preference for a supply chain with fewer places to hide.

Knowing where responsibility formally sits does not, by itself, prevent the dispute. It changes what you do during one. A business that knows its contract holder is the accountable party stops trying to adjudicate between suppliers and starts doing something more effective: holding one company to an obligation it has already accepted. The conversation shifts from “can you help me work out whose fault this is” to “you are contractually responsible for restoring this service, so please tell me what you are doing and when the next update is due.” Those are very different calls, and they produce very different outcomes, from exactly the same contractual position.

The reason so many businesses never make the second call is that they are genuinely unsure who the contract holder is. UK SME connectivity is frequently bought through an intermediary - a managed IT provider, a telecoms reseller, an office equipment supplier, sometimes a landlord or serviced-office operator. The invoice may come from one company, the technical support from another, the router from a third and the physical line from a fourth. Each of those relationships is legitimate on its own. Stacked together without a documented escalation path, they produce exactly the ambiguity that 34% of buyers report.

The layers in a typical UK connectivity supply chain

The chain that produces these disputes is not exotic. It is the standard shape of the UK market, and most businesses sit inside a version of it without having mapped it. At its most common there are three or four distinct parties between the business and the fibre.

At the base sits the infrastructure owner - the organisation that physically owns the ducts, poles, fibre and exchange equipment. In much of the country that is the incumbent national network; in a growing number of areas it is an alternative network operator, or altnet, that has built its own fibre. Above that sits the wholesale operator that sells access to that infrastructure to service providers. Above that sits the retail internet service provider that packages the access, adds routing, support and billing, and sells it to businesses. And in many SME arrangements there is a fourth layer: a managed IT support provider or reseller that sells the connectivity alongside its other services, and which is very often the only party the business actually speaks to day to day.

Every one of those layers is a legitimate commercial function. The problem is not the existence of the chain, it is that fault ownership is rarely defined across it in writing. Each layer typically has a clear obligation to the layer directly above it, and none has an obligation to the business at the top except the one it directly contracts with. When a fault sits in a boundary - not clearly inside any single layer’s domain - each party can test its own segment, find it healthy, and hand the fault back in good faith. Nobody is behaving badly. The chain is simply doing what an undefined chain does.

Read that grid as a diagnostic rather than a scorecard. The four high-severity items share a characteristic: each is a gap that only becomes visible during an incident, which is the worst possible moment to discover it. The mid-severity items are the ones that produce a false sense of protection - a backup circuit that has never been tested, or a diverse route that turns out not to be diverse, are arguably more dangerous than having no backup at all, because they cause a business to skip the contingency planning it would otherwise have done.

What a disputed outage costs a UK business

The survey does not price these incidents, and any single national figure for the cost of downtime is close to meaningless because the variation between organisations is enormous. What is useful is a method. The table below is an illustrative model, not survey data: it shows how the cost of a lost working day scales across typical UK SME size bands, using a simple and conservative calculation - staff time made unproductive plus trading revenue that cannot be transacted while systems are unreachable. Substitute your own headcount, average loaded staff cost and daily revenue to get a figure that means something for your organisation.

The final column is the one that changes how a business should think about connectivity procurement. Service credits are not compensation for business loss; they are a partial refund of line rental, calculated as a proportion of a monthly charge and usually capped. No realistic SLA in the UK SME market makes a business whole after a lost trading day, and none is designed to. That means the commercial protection against a disputed outage cannot come from the credit regime. It has to come from the design of the arrangement - who is accountable, how quickly they must respond, how the fault escalates, and what happens to traffic while the primary circuit is down.

It also reframes the price-versus-response finding. If a provider with clear single-point accountability costs an extra £40 a month - £480 a year - and it removes even one four-hour disputed interval from a small business over the life of a three-year contract, the arithmetic is not close. That is why 56% of buyers now cite outage response ahead of price. They are not being sentimental about service quality. They are pricing an asymmetric risk correctly.

Reactive versus accountable: two ways to hold a connectivity supply chain

The distinction between the two columns is not budget. Almost nothing in the right-hand column is expensive in isolation - documenting a contract holder costs an afternoon, a monitoring probe on a circuit costs very little, requiring root cause analysis is a clause rather than a capital item. What separates the columns is whether the work was done before the incident or attempted during it. Every item in the reactive column is the same item from the accountable column, attempted under time pressure by people who are also trying to keep a business running.

Why cloud and AI have raised the cost of ambiguity

75% of the buyers surveyed say cloud and AI adoption has made resilient, always-on connectivity more important to their operations than it was before. That single figure explains why a structural problem the industry has lived with for years is now surfacing as a commercial one. The connectivity supply chain has not become more tangled recently. What has changed is how much a business loses per hour when the chain fails to resolve a fault.

Consider what a typical UK SME had on-premises in the recent past and what it has now. Email ran on a local server; it now runs in Microsoft 365. Files sat on a network share; they now sit in SharePoint or a cloud storage platform. The phone system was a physical PBX in a cupboard; it is now a hosted VoIP service, and the analogue network it used to fall back on is being switched off. Accounting was a desktop application; it is a browser tab. Card payments ran over a dedicated line; they run over the internet connection. Backup ran to tape or a local appliance; it replicates to cloud storage. Each of those migrations was individually sensible and individually cost-reducing. Collectively they moved every critical dependency in the business onto one circuit and removed every offline fallback that used to absorb an outage.

AI adoption extends the same trend one step further, and in a way that is less obvious. AI-assisted tools are not merely cloud-hosted; they are cloud-hosted, latency-sensitive and used interactively. A document assistant, a transcription service, a customer-facing assistant or a coding tool degrades badly on a marginal connection rather than failing cleanly. That matters for fault ownership, because a degraded connection is precisely the kind of fault most likely to be disputed. A hard circuit failure is unambiguous and testable; every party in the chain agrees the line is down. Intermittent packet loss, elevated latency at certain hours or a jitter problem affecting voice and interactive AI traffic will pass most line tests. It is the class of fault where each layer can honestly report that its segment is within specification, and the class most likely to bounce between suppliers for days.

That is the practical link between the 75% and the 45%. Increasing dependency on cloud and AI does not just raise the cost of outages - it increases the proportion of faults that are subtle, boundary-spanning and therefore contestable. A business that has moved everything to the cloud has more to lose from exactly the type of fault its supply chain is worst at owning.

The end-to-end network argument, and where it stops

61% of buyers say they prefer providers with an end-to-end, owned and operated network, and 66% say infrastructure ownership matters more than it did a year ago. The logic is sound: fewer parties means fewer boundaries, and boundaries are where faults get disputed. A provider that owns the fibre, the network and the customer relationship cannot hand a fault to a third party, because there is no third party. When something breaks, the company answering the phone is the company that can fix it.

It is worth being precise about the limits of that argument, because it is being made by a company with a commercial interest in it. Owning the network removes the wholesale boundary; it does not remove every boundary. Faults still occur in the customer’s own equipment, in the router configuration, in the local wiring, in third-party works damaging a duct, in power, and in the applications riding the circuit. An owned-and-operated network also has a geographic footprint - it can only serve a business at an address it has built to, and coverage varies enormously across the UK. And ownership says nothing on its own about support quality, out-of-hours staffing or escalation discipline, which are the operational behaviours the 56% switching figure is really about.

The defensible version of the finding is narrower and more useful than the marketing version: what buyers want is a shorter accountability chain and a clear owner. Vertical integration is one way to achieve that. A well-structured managed arrangement, where a single provider takes contractual ownership of the whole stack and handles the supplier-to-supplier conversations on the customer’s behalf, is another. What both have in common is that the business never becomes the integrator. Whether the layers underneath belong to one company or four matters far less than whether the business at the top has to manage them.

What to establish before you sign or renew

Connectivity contracts are typically signed for 24 or 36 months, which means most UK businesses touch this decision only rarely - and usually under mild time pressure at the end of a term. The questions below are the ones the survey findings point at directly. They are not technical questions; they are accountability questions, and every one of them should be answerable in writing before signature rather than discovered during an incident.

Ask who holds the contract and who owns the infrastructure underneath it, by name, and get both answers written into the order documentation. Ask whether the SLA clock starts at loss of service or at fault acceptance - this is the single most consequential clause in most connectivity contracts and the one buyers most consistently overlook. Ask what the escalation path is by role and timescale, not by individual, so that it survives staff turnover on both sides. Ask whether proactive monitoring is included and whether it generates an outbound alert to you, since the alternative is that your staff are the monitoring system. Ask whether a root cause analysis is provided as standard after a P1 incident, and whether it is contractual or discretionary. Ask what happens to your traffic during a primary circuit failure, and whether any backup path is genuinely physically diverse rather than a second product riding the same duct and the same exchange. And ask who is responsible if the fault turns out to sit at a boundary - between the router and the line, between the line and the wholesale network, between the network and an application. That last question is the one that separates a supplier who has thought about the 45% from one who has not, and the quality of the answer tells you more than any speed or price comparison will.

At a glance

How this fits the wider connectivity picture

This finding does not sit in isolation. It is the third strand of a pattern that has run through UK business connectivity reporting all year. In our coverage of the UK’s mobile performance gap and the AI traffic warning, operators argued that rising AI-driven demand will make today’s occasional congestion routine within two to three years - the capacity side of the same dependency this survey measures on the accountability side. The consolidation story we examined in the Gamma and Epiris buyout and what it means for VoIP continuity deals with the ownership question directly: when the company behind your hosted phone system changes hands, the identity of your accountable party changes with it. And the NCSC guidance on internet-exposed edge devices covers the equipment sitting at the boundary of exactly the chain described here - the routers and firewalls that are, in practice, the most commonly disputed piece of kit in an outage.

Two further pieces are worth reading alongside this one for the governance angle. The UK airports breach and its penetration testing lesson makes the same structural argument in a security context: shared responsibility across a supplier chain fails at the boundaries unless somebody has been named as accountable and the assumption has been tested. And the Claude Code prompt injection exploit illustrates how quickly a new class of dependency arrives inside a business before anybody has written down who owns it - the same lag that produced the 34% figure in this research.

Frequently asked questions

CloudSwitched

London-based managed IT services provider offering support, cloud solutions and cybersecurity for SMEs.