Choosing a VoIP Provider: A UK Business Guide to Comparing Hosted PBX Options Beyond Price in 2026

Serious VoIP provider comparison starts with the things that are hardest to reverse, not the number on the front of the quote. Price per seat is the easiest column in the spreadsheet to fill in and the least predictive of how the next four years will actually feel. The per-user figure is set by a sales team that wants the signature; the contract term, the porting process, the support model and the integration surface are set by the provider’s underlying architecture, and none of those change because you negotiated a pound off the licence.

This guide is written for whoever has been handed the job of replacing a phone system and has three quotes on the desk that all look broadly similar. It works through what genuinely differentiates one hosted PBX from another — who actually owns the platform, where the numbers live, what the service level agreement commits to in writing, how long a port really takes and what happens to your estate at the end of the term. It covers the contract clauses that quietly decide whether you can leave in year three, the integration questions that decide whether the system fits the way your organisation works, and a costed picture of UK 2026 pricing that includes the lines most quotes leave out. The aim is that by the end you can score a shortlist on something more durable than hosted PBX selection by headline price.

What you are actually buying when you buy hosted voice

A hosted PBX is a telephone system that runs in a provider’s data centres rather than in a cupboard in your building. Handsets, softphones and mobile apps register across the internet to that platform; the platform holds the dial plan, the call routing, the voicemail, the call recording and the reporting, and it connects out to the wider telephone network through interconnects with other carriers. You pay a monthly licence per user, usually with a bundle of minutes, and the provider handles the upgrades. That much is common to every quote on your desk, which is precisely why the quotes look interchangeable.

The differences sit one layer down, and the single most useful question in any SIP trunk provider evaluation or hosted PBX shortlist is this: does the company selling to you actually operate the platform, or are they reselling someone else’s? The UK market has perhaps a dozen genuine platform operators — the likes of Gamma, 8x8, RingCentral, Zoom, Vonage, Microsoft with Teams Phone, and a handful of wholesale carriers — and several thousand resellers, dealers and white-label partners selling those platforms under their own branding. Neither model is wrong. A good reseller with real engineering depth will out-support a direct account at a large operator every time, because you are a meaningful customer to them and a line item to the operator. But you need to know which one you are buying, because it changes who fixes a fault at 4pm on a Friday, who holds your numbers, and what happens if the middle party goes out of business.

The second structural question is where your telephone numbers sit in the chain. In the UK, number ranges are allocated by Ofcom to communications providers, and a provider that holds ranges in its own right is described as the range holder. If your provider is a reseller, your numbers are almost certainly allocated on the underlying operator’s ranges, sub-allocated to you. That arrangement is entirely normal and works fine day to day, but it adds a party to every porting conversation and, in the worst case, means that a dispute between your reseller and their wholesaler can sit between you and your own phone numbers. Ask the question during procurement, in writing, and note the answer. It costs nothing to ask and it is a genuinely awkward thing to discover later.

The third is the relationship between the voice service and the connection it rides on. Some providers sell you both and will therefore accept end-to-end responsibility when call quality degrades; others sell voice only and will, reasonably enough, point at your broadband when jitter climbs. Neither answer is disqualifying, but a single-supplier model removes an argument you would otherwise have to have during an outage, and a split model usually costs less and gives you leverage. Decide which you want before the sales calls start rather than discovering your preference halfway through a fault.

Direct from the platform operator, or through a reseller?

This is the first fork in a VoIP provider comparison and the one most buyers never consciously make, because the reseller’s quote and the operator’s quote arrive in the same inbox looking like the same product. They are the same product, technically. What differs is the commercial and support structure wrapped around it, and that structure is what you experience for the whole of the term.

Buying direct from a platform operator gives you the shortest fault path, first sight of platform changes and, generally, the better price at scale. It also puts you into a support model built for volume: tiered ticket queues, published response targets, and an account manager whose patch might run to several hundred customers. Buying through a reseller puts a named engineering team between you and the platform. At their best, they know your dial plan, they answer the phone, and they absorb the tedious work of chasing the wholesaler. At their worst, they are a billing intermediary with a partner portal and no deeper knowledge of the platform than you have.

The premium attached to the partner route is real but usually modest — commonly one to three pounds per user per month at SME scale, sometimes nothing at all where the partner’s wholesale rate is better than the operator’s list. What that premium buys is triage. When a handset stops registering, somebody who already knows your configuration looks at it before a ticket is raised with the operator, and in the large majority of cases the fault never leaves the partner. Whether that is worth paying for depends entirely on whether you have anyone in-house who would otherwise do it.

Testing a partner’s depth is straightforward and worth the half hour. Ask them to explain, unprompted, how they would handle a one-way audio fault on a single site. A genuine engineering partner will talk about NAT traversal, SIP ALG on the firewall, RTP port ranges and codec negotiation, and will want to know what router you have. A billing intermediary will say they would raise it with the platform. Both answers are honest; only one of them tells you the fault will be fixed the same afternoon. The same logic applies to the connectivity underneath — the discipline described in our guide to bandwidth planning for a growing UK business is exactly what a competent partner brings to a voice deployment, and its absence is what turns a clean install into a quality complaint.

Where hosted PBX due diligence usually falls short

Procurement for a phone system tends to be thorough about the things that are easy to check and silent about the things that matter. Handset models, minute bundles and per-seat price get compared in forensic detail. Porting exposure, exit mechanics and integration compatibility get a nod in a meeting and never appear in the evaluation document. The grid below reflects the pattern we see repeatedly when a business asks for help unpicking a contract it signed two or three years earlier: the high-risk items are almost never the ones that were scrutinised at the time.

The item worth pausing on is “free handsets on a longer term”, which appears in the over-weighted column carrying a high risk badge. Free hardware is rarely free; it is hardware amortised across a longer minimum term, and in a meaningful minority of UK deals it is financed through a separate lease agreement with a third-party funder. That second agreement has its own term, its own termination provisions and its own auto-renewal, and cancelling the voice service does not cancel it. More than one organisation has moved provider cleanly and continued paying for handsets it no longer uses for another two years. If hardware is described as free or heavily discounted, ask directly whether a finance or lease agreement forms part of the paperwork, and ask to see it before anything is signed.

The UK hosted voice market in numbers

Some context for the decision. The UK business telephony market is in the final stage of a structural migration off copper, which means an unusually large number of organisations are buying a phone system at the same time, many of them for the first time in a decade. That has two practical effects: provider sales teams are busy and discounting hard, and porting queues at the wholesale level are longer than they were. Both of those facts should shape how you run the procurement and how much time you leave for the cutover.

The spread on price deserves a comment, because it is wider than most comparison sites suggest. The bottom of the range — around six to eight pounds per user per month — buys a functional hosted extension with a softphone, voicemail and basic call routing, typically from a high-volume provider with self-service support. The top of the range buys a full unified communications platform with contact-centre features, analytics, native Microsoft Teams integration and a managed support wrap. Both are honest prices for what they are. Comparing them as though they were the same product is the single most common error in business phone system procurement, and it is the reason so many organisations end up disappointed by a service that is doing exactly what they paid for.

The term length matters more than the monthly figure in most cases. A three-pound-per-seat saving on forty users is £120 a month; over a five-year term that is £7,200, which sounds decisive until you consider that a five-year term also means you are committed to a 2026 platform in 2031, and that the industry has changed materially in every preceding five-year window. Where possible, take the shorter term at the slightly higher price. The option value of being able to leave is worth more than the discount, and it is the one thing you cannot buy back later.

Why UK businesses actually change VoIP provider

The most instructive evidence in a VoIP provider comparison is not what buyers say they want at the start; it is what makes them leave. When organisations move from one hosted platform to another, price is rarely the trigger. The pattern below reflects the reasons that come up repeatedly in switching conversations, weighted by how often each is named as the primary driver rather than a contributing grumble.

Read that chart against the way most evaluations are scored and the mismatch is obvious. Support quality is the dominant reason organisations leave, and it is the hardest thing to assess from a proposal document. Every provider claims responsive UK-based support; the claim is free to make and nearly impossible to falsify in a sales meeting. The only reliable tests are indirect ones: reference customers of a similar size in a similar sector, published response targets with financial consequences attached, and a trial of the support channel itself before signing.

That last one is underused and genuinely revealing. During the evaluation, raise a low-priority ticket with each shortlisted provider — a configuration question, a request for a portal walkthrough — and time the response. You are not testing whether they can answer; you are testing what the queue feels like when you are a prospect they want. Whatever that experience is, it is the best it will ever be. If a provider takes two days to answer a pre-sales question, the post-sales reality will not be quicker.

Call quality sitting second is worth unpacking too, because responsibility for it is genuinely shared. A hosted platform cannot deliver clean audio across a congested or poorly configured connection, and a large proportion of quality complaints resolve to the customer’s own network rather than the provider’s platform — SIP ALG left enabled on a small-business router, no traffic prioritisation, a contended connection at 4pm, or a wireless network carrying handsets it was never designed for. The way to keep this from becoming an unresolvable argument is to agree, before installation, what the network baseline is and who verifies it. Continuous visibility helps here as well; the approach set out in our guide to proactive network monitoring for UK businesses turns “the phones were bad on Tuesday” into a timestamped jitter graph that settles the question in minutes rather than weeks.

How to weight a scoring matrix that reflects reality

If price per seat is the wrong primary criterion, something has to replace it. The weightings below are a defensible starting point for a mid-sized UK organisation buying hosted voice in 2026. They are not universal — a contact centre weights integration and analytics far higher, a two-site professional services firm weights resilience lower — but they are a better default than the price-dominant scoring most procurement templates ship with. Treat the percentages as the share of the total decision each factor should carry.

Notice that cost still carries fourteen per cent — this is not an argument for ignoring price. It is an argument for measuring the right price. “Total cost over the full term” means the licence, the minutes you will actually use, the hardware, the installation and professional services, the porting charges, any integration licensing, and the cost of the connectivity if it is being upgraded to support voice. That figure is frequently forty to sixty per cent higher than the per-seat quote multiplied by headcount and term, and it is the only cost number worth putting in front of a board.

Integration fit at thirteen per cent is the weighting most often set too low. A phone system that does not connect to the CRM produces an organisation where nobody logs calls, which means the CRM data degrades, which means the reporting that justified the CRM stops being trustworthy. The cost of that is real but arrives slowly and never gets attributed to the telephony decision. It is the same category of compounding data problem covered in our guide to reporting databases and data warehouses for UK businesses — the damage is not the missing integration, it is the years of incomplete records that follow it.

Security and regulatory posture at eight per cent is deliberately modest but not negligible. For most organisations it means confirming three things: that the provider can demonstrate a recognised security standard such as ISO 27001 or Cyber Essentials Plus, that call recordings are stored within a jurisdiction compatible with your UK GDPR obligations and are exportable on request, and that fraud controls exist on outbound calling. That third point is routinely overlooked until a compromised extension dials premium-rate international destinations overnight and leaves a five-figure bill that the contract makes the customer’s responsibility. Ask what call barring, spend caps and anomaly alerting are enabled by default, and what the provider’s liability position is if fraud occurs.

The real procurement and migration timeline

Almost every phone system project runs late for the same reason: the organisation budgets time for the decision and no time for the port. Choosing a provider takes weeks and is entirely within your control. Porting numbers takes weeks, involves at least two other companies and a regulated process, and is not within your control at all. The timeline below reflects a realistic sequence for a thirty-to-eighty user organisation with a handful of main numbers and a DDI range.

The headline number in that timeline is fourteen weeks from first inventory to cutover, and eighteen to a closed project. Providers routinely quote four to six. Both can be true: the provider’s number describes their build work, yours has to describe the whole programme including the regulated porting process and the decisions that precede it. Plan against the longer figure and treat anything faster as a bonus, particularly in the run-up to the PSTN withdrawal when porting queues are at their busiest. Our guide to the PSTN switch-off and VoIP migration covers the copper-specific dependencies in detail if you are still running analogue or ISDN services.

A readiness gauge for your own procurement

Before comparing providers, it is worth scoring your own preparation, because the quality of a VoIP provider comparison is limited by the quality of the requirement behind it. Organisations that can answer the ten questions below score well and tend to run short, decisive procurements. Organisations that cannot tend to be led by whichever sales team is most attentive, which is not the same as whichever provider is most suitable.

Score one point for each of the following that you can answer today, from documented evidence rather than recollection, then multiply by ten. Most organisations we meet score four or five before any preparation work, which is why the benchmark above sits where it does.

• How many telephone numbers does the organisation hold, and which provider bills each one?
• What is the peak number of simultaneous calls across the business, from actual call records?
• Which non-telephone devices depend on a line — lift phones, alarm signalling, door entry, card terminals?
• What does the current service cost in total, including line rental, call spend, maintenance and hardware finance?
• When does the existing contract end, and what notice period applies?
• Which business systems must the phone system connect to, and who owns each of those systems?
• What is the upload capacity and current utilisation of the connection at each site?
• Who answers the phone, in what order, and what happens to a call that nobody picks up?
• What are the out-of-hours, holiday and emergency routing rules, and where are they documented?
• Is call recording required for compliance, and if so what retention period applies under your UK GDPR obligations?

The exercise takes a day or two and it changes the procurement entirely. A provider responding to a documented requirement quotes against your business; a provider responding to “we have about forty staff and need new phones” quotes against their standard package and fills the gaps with assumptions you will discover during installation. It also gives you a defensible basis for comparing quotes that are otherwise structured differently on purpose. Fitting the telephony decision into a broader plan helps too — the sequencing discipline in our guide to building an IT roadmap and technology strategy keeps a voice project aligned with the connectivity and identity work that usually surrounds it.

UK hosted PBX pricing in 2026 — what each tier actually includes

The table below sets out the realistic bands in the UK market in 2026 and, more usefully, what changes between them. The per-user figures are typical list prices before negotiation; expect ten to twenty per cent movement on a competitive three-year deal and more at higher seat counts. The right-hand column is the one to read carefully, because it describes what you are giving up at the cheaper tiers.

Three cost lines sit outside that table and belong in any honest total. The first is professional services: build, configuration, porting coordination and training, typically £500 to £3,500 depending on complexity, sometimes waived on a longer term — which, as with free handsets, is a discount you repay through the term length. The second is hardware, at roughly £70 to £120 for a basic desk phone, £130 to £220 for a mid-range colour handset with expansion capability, and £180 to £400 for conference units and DECT solutions. The third is connectivity, where a business moving from copper to a dedicated or higher-capacity circuit can add £150 to £600 a month, an amount that has nothing to do with the voice provider but everything to do with the project budget.

Call charges deserve one specific warning. Bundles are normally quoted as inclusive UK landline and mobile minutes, which covers the great majority of traffic for most organisations. What sits outside the bundle is international, non-geographic ranges, premium-rate destinations and, on some tariffs, calls to 070 personal numbers. Those out-of-bundle rates vary by a factor of ten or more between providers and are almost never compared during evaluation. If your business calls internationally with any regularity, pull the destination breakdown from three months of bills and price it explicitly against each quote. It is common for the provider with the higher seat price to be cheaper in total once real traffic is applied.

How much of the decision is still made on price alone

For all the material available on evaluating communications suppliers, the dominant selection criterion in the UK SME market remains the monthly figure per seat. That is not irrational — it is the only number that is directly comparable across three proposals written to three different templates — but it is a proxy for value rather than a measure of it, and it systematically favours providers who have moved cost out of the visible line and into the contract.

The consequence shows up two to three years later rather than immediately, which is why the pattern persists. A provider competing purely on the visible number has three levers: lengthen the term, shift cost into hardware finance, and reduce the support wrap. All three are invisible in a side-by-side price comparison and all three are exactly what the organisation complains about later. None of this requires bad faith on the provider’s part; it is a rational response to being evaluated on one column.

The correction is not to ignore price but to compare the right quantity. Build a four-year total for each shortlisted provider covering licences, realistic call spend from actual bills, hardware whether purchased or financed, professional services, porting charges and any connectivity uplift. Then apply a stated assumption about headcount change — most organisations are not the same size in four years as they are today — and check what each contract does when the number of users goes up or down. A contract that allows seats to be added at the agreed rate but never reduced is a materially different commercial proposition from one that flexes both ways, and that difference will not appear anywhere in a per-seat comparison.

The twelve-point contract and SLA checklist before you sign

This is the part of the process where a couple of hours of attention is worth more than any amount of negotiation on price. Every item below is something we have seen cause real difficulty for a UK business after signature, and every one of them is easier to fix before the contract is executed than at any point afterwards. Work through them in order with the agreement in front of you.

1. Minimum term and what starts it. Confirm the term length in months and, critically, the date it starts — signature, service activation or porting completion. A term that starts at signature while porting takes three months quietly costs you a quarter of paid service you could not use.
2. Renewal mechanism and notice period. Establish whether the agreement rolls into a further fixed term automatically or converts to a rolling monthly arrangement. Auto-renewal into another full term with a ninety-day notice window is the single most common lock-in trap in UK business telecoms. Set a calendar reminder for the notice date on the day you sign.
3. Price review clause. Find out whether prices can rise mid-term and on what basis. Inflation-linked or index-plus-a-margin increases are still common in business agreements even where consumer rules have tightened. Ask for a fixed price for the term, or a stated cap; providers frequently agree when asked and never offer when not.
4. Early termination charge basis. Determine whether the charge is the full remaining contract value, a percentage of it, or a defined number of months. Full remaining value on a five-year term is not a termination clause in any practical sense; it is a prohibition.
5. Hardware ownership and any finance agreement. Establish who owns the handsets at the end of the term and whether a separate lease or finance agreement exists with a third-party funder. If it does, read it separately — it has its own term, its own notice period and its own auto-renewal, none of which are cancelled by ending the voice service.
6. Number ownership and porting-out cooperation. Confirm in writing that you can port your numbers away at the end of the term, what the provider charges to release them, and what the committed timescale is. A provider unwilling to put porting-out cooperation in the contract is telling you something useful.
7. Service level agreement with consequences. A published uptime target with no service credit attached is a marketing statement. Look for a defined availability commitment, defined fault priorities, target response and restoration times per priority, and a credit mechanism that applies automatically rather than on application.
8. Support hours, channels and escalation path. Get the actual hours of technical support as distinct from account management, the channels available at each priority, and the named escalation route with timescales. Confirm whether out-of-hours cover is included or chargeable, and where the support team is located.
9. Scaling in both directions. Check the mechanism and pricing for adding seats and, separately, for removing them. Many agreements permit unlimited growth at the contracted rate and prohibit any reduction below the committed number for the whole term. If your headcount is uncertain, negotiate a permitted reduction band of ten or fifteen per cent.
10. Data, recordings and exit. Establish where call recordings and call detail records are stored, how long they are retained, whether retention is configurable to match your UK GDPR obligations, and in what format and timescale you can export everything on exit. Getting three years of recordings out of a platform you are leaving is considerably harder than it sounds.
11. Fraud liability and outbound controls. Confirm what call barring, destination blocking and spend alerting are enabled by default, whether a hard spend cap is available, and who bears the cost if an extension is compromised. Assume the contract places it with you unless it says otherwise, and configure the controls on day one.
12. Emergency calling and registered addresses. Verify how 999 and 112 calls are routed, what address is presented to the emergency operator for each number, how that address is updated when someone moves desk or site, and what happens to emergency calling during a power cut or internet outage. Regulatory obligations sit with the provider; keeping the address data accurate sits with you.

One further note on the service level agreement. Uptime figures in this market are quoted with a confidence that rarely survives arithmetic. A 99.9 per cent availability commitment permits roughly forty-three minutes of downtime a month, or about nine hours a year. A 99.99 per cent commitment permits around four minutes a month. The difference between those two figures is the difference between an inconvenience and an incident, and providers quote both as though they were near-identical. Establish which number is committed, whether it covers the whole service or only the core platform, and whether planned maintenance is excluded from the calculation — it usually is.

Integration compatibility — the question behind the tick box

Every provider’s comparison sheet has a row marked “CRM integration” with a tick in it. The tick is almost always defensible and almost never sufficient, because it covers at least four quite different things. Knowing which one you are being offered is the difference between a system people use and a system people work around.

The shallowest form is click-to-dial, usually delivered by a browser extension that turns telephone numbers on a web page into clickable links. It is genuinely useful, takes ten minutes to deploy, and tells you nothing about whether calls are recorded against records. The next level up is screen pop, where an inbound call triggers a lookup against the CRM and opens the matching contact. This requires a real integration with the CRM’s API and is where implementations start to differ in quality — whether it matches on multiple number formats, whether it handles calls from a number held against several records, and how it behaves when there is no match.

Third is call logging, where the system writes an activity record against the contact automatically, with direction, duration, outcome and a link to any recording. This is the one that matters for reporting, and it is the one most commonly missing from an implementation that was sold as integrated. Fourth is bidirectional workflow, where the CRM can drive telephony actions and telephony events can trigger CRM automation. Only a minority of organisations need this, but those that do should establish it explicitly, because it usually requires API access that sits on a higher licence tier.

Ask which of those four you are getting, for which specific version of which specific system, and whether the connector is built and supported by the telephony provider, by the CRM vendor, or by a third party. That last distinction determines who fixes it when an update breaks it, and updates do break it. A third-party connector with a single maintainer is a dependency worth knowing about before it becomes a fault nobody owns.

The same discipline applies to Microsoft Teams, which is where a large share of UK integration questions now land. There are three broad routes and they are not equivalent. Direct Routing connects a SIP trunk or provider to Teams through a certified session border controller, keeps Teams as the client and typically offers the widest carrier choice. Operator Connect places the carrier inside the Teams admin experience with a simpler onboarding path and a shorter list of participating operators. A provider’s own softphone running alongside Teams is not Teams integration at all, though it is frequently presented as such — it gives users two applications that both ring. Establish which route is on offer, what it requires in Microsoft licensing, and who owns the session border controller if one is involved. The licensing is the part that most often changes the total cost after the decision has been made.

Finally, check the less glamorous integrations that quietly matter: directory synchronisation with Microsoft Entra ID or Google Workspace so that starters and leavers are handled once rather than twice, single sign-on so that the telephony portal is inside your existing identity controls, and any connection to a service desk or practice management system. Identity integration in particular is a security control, not a convenience — a phone system with its own separate user database is an account set that survives offboarding, which is the kind of gap an assessment like a Cyber Essentials Plus technical audit is designed to surface.

A real-world example — when the cheapest quote cost the most

A forty-two-person architectural practice with offices in Leeds and Bristol replaced an ageing on-premises PBX in early 2024. Three quotes were received. The practice selected the lowest, at £7.50 per user per month against £11.20 and £12.90, on a sixty-month term with handsets provided at no upfront cost. On the visible arithmetic the saving looked substantial: roughly £185 a month against the middle quote, a little over £11,000 across the term, plus around £5,000 of handsets they did not have to buy.

Three things emerged over the following eighteen months. The handsets were supplied under a separate lease with a third-party funder on a sixty-month term of its own, at £103 a month — a cost nobody had added to the comparison because it had been presented as included. The tariff carried an annual index-linked increase, which added a little over nine per cent across the first two reviews. And the support model was portal-only during business hours, which the practice discovered during a one-way audio fault affecting the Bristol office that took eleven days to resolve because the provider could not attend site and the practice had no partner who could.

By the time the total was recalculated at the eighteen-month mark, the “cheapest” option was running approximately £90 a month above what the middle quote would have cost, before any value was assigned to the eleven days of degraded service or the internal time spent chasing it. The practice wanted to move. The early termination charge was the full remaining contract value on both the service agreement and the hardware lease, which made moving before the end of the term commercially impossible. They completed the full five years.

We spent six weeks comparing feature lists and about twenty minutes on the contract. The feature lists were all basically the same. The contract was the entire decision and we did not read it properly. If I did it again I would start with the exit terms and work backwards.

The instructive detail is that nobody involved behaved improperly. The provider quoted accurately for what it offered, the lease agreement was disclosed in the pack, and the index-linked clause was in the terms. The failure was one of evaluation method: a comparison built around a single visible number, against three proposals deliberately structured so that the visible number was the only comparable element. The practice has since moved to a thirty-six month agreement with a fixed price for the term and a defined reduction band, at a higher per-seat rate.

Common mistakes in VoIP provider selection

Most procurement failures in this category are variations on a small number of themes. If you recognise more than two of the following in your current process, it is worth pausing before the shortlist is finalised.

• Comparing per-seat prices across different product tiers. An entry hosted extension and a full unified communications licence are not competing offers, and placing them in the same column makes the cheaper one look like better value rather than a different product. Normalise the tier before comparing the price.
• Cancelling the existing service before the port completes. This is the most damaging error available in the whole process. Once the losing provider ceases the line, the number is generally gone and cannot be recovered. Keep the old service live and paid for until every number has ported and been tested.
• Submitting porting data from memory rather than the bill. Port requests are rejected on mismatched account numbers, postcodes and account names, and each rejection restarts the clock. Copy the details from the current provider’s bill exactly as printed, including how the company name is spelled on the account.
• Treating the network as somebody else’s problem. Call quality is produced jointly by the platform and your own connection, switching and firewall. A survey and the necessary remediation before installation prevents the majority of quality complaints, and prevents the months of unresolvable finger-pointing that follow when it is skipped.
• Forgetting the devices that are not telephones. Lift emergency phones, intruder alarm signalling, door entry panels, card payment terminals and postal franking machines all use lines and none of them appear in a headcount. Each needs its own migration route, and the lift line in particular carries a statutory safety obligation that does not pause for a telephony project.
• Accepting an uptime figure without the consequence. A stated availability percentage with no service credit, no defined fault priorities and no restoration targets is not a service level agreement. Ask what happens commercially when the target is missed; if the answer is nothing, the target is decorative.
• Buying for today’s headcount on a five-year term. Organisations grow, shrink and reorganise. A contract with no reduction mechanism converts a normal change in circumstances into a fixed cost you carry for years. Negotiate a permitted reduction band at the point of signature.
• Skipping reference calls with similar organisations. Case studies are curated; reference calls are not. Two twenty-minute conversations with customers of comparable size in a comparable sector will tell you more about the support experience than any amount of proposal reading.

At a glance — the VoIP provider comparison summary

The key facts from this guide in one place, for pasting into an evaluation document or a board paper.

How Cloudswitched approaches hosted voice

Cloudswitched delivers hosted VoIP and SIP trunking for UK businesses as a managed service: line and number audit, requirements documentation, network readiness survey and remediation, platform build, number porting coordination, cutover support and ongoing management of the estate. Where an organisation prefers to buy direct from a platform operator, we will say so and help structure the evaluation rather than compete for the sake of it — the point of the exercise is a system that fits, on terms you can live with for the length of the contract.

Frequently Asked Questions

How much should a UK business expect to pay per user for hosted VoIP in 2026?

The realistic range is £6 to £25 per user per month, and the spread reflects genuinely different products rather than different margins. Around £6 to £9 buys an entry hosted extension with a softphone, voicemail and basic routing, usually with portal-only support. Ten to fourteen pounds buys a standard business tier with desk phone provisioning, multi-level IVR, queues and on-demand recording. Fifteen to twenty-two pounds buys full unified communications with video, a native Microsoft Teams route, CRM connectors and analytics. Contact-centre licensing starts around £25 and runs to £65 or more. Before comparing any of these figures, confirm which tier each quote represents, because comparing an entry extension with a UC licence makes the cheaper one look like better value when it is simply a smaller product.

How long does it take to port business phone numbers to a new VoIP provider?

A simple single-line port typically completes in around ten working days from a clean submission. A multi-line port, or one involving DDI ranges, commonly takes fifteen to twenty-five working days, and complex estates spanning several providers can take longer. The critical detail is that any rejection restarts the clock, and the most frequent cause of rejection is a mismatch between the details on the porting request and the details held on the losing provider’s account — account number, postcode, or the exact spelling of the company name. Copy those details from a current bill rather than from memory. Plan the project against twenty-five working days rather than ten, and never cease the existing service until every ported number has been tested inbound and outbound.

Is it better to buy VoIP direct from the platform operator or through a reseller?

It depends on whether you have anyone in-house who can triage a telephony fault. Buying direct from a platform operator gives the shortest fault path and usually the better price at scale, but places you in a support model built for volume. Buying through a managed partner typically adds one to three pounds per user per month and puts a named engineering team between you and the platform, which matters most for organisations of roughly five to a hundred and fifty seats without internal voice expertise. The test that separates a capable partner from a billing intermediary is to ask how they would approach a one-way audio fault; a real engineering partner will discuss NAT traversal, SIP ALG and codec negotiation, while an intermediary will say they would raise it with the platform.

What should a VoIP service level agreement actually contain?

A usable agreement contains four things: a defined availability commitment, a fault priority scheme, target response and restoration times for each priority, and a service credit mechanism that applies when targets are missed. A published uptime percentage with none of the other three is a marketing statement rather than a commitment. Check the arithmetic as well — 99.9 per cent availability permits roughly forty-three minutes of downtime per month, while 99.99 per cent permits about four, and providers quote both as though they were equivalent. Establish whether the figure covers the whole service or only the core platform, and whether planned maintenance is excluded from the calculation, which it usually is.

What contract terms cause the most problems for UK businesses?

Four clauses account for most of the difficulty. Automatic renewal into a further fixed term, often with a ninety-day notice window, catches organisations that assumed the contract would become rolling. Early termination charges calculated as the full remaining contract value make leaving before the end of the term commercially impossible on a long agreement. Mid-term price review clauses linked to an inflation index compound quietly across the years. And separate hardware finance agreements with third-party funders continue running after the voice service is cancelled. Each of these is negotiable before signature and effectively fixed afterwards, so the contract review deserves more time than the feature comparison.

Can we keep our existing telephone numbers when we change provider?

In almost all cases yes — UK number portability is a regulated obligation and geographic numbers can generally be ported between communications providers. The practical risks are procedural rather than legal. If the existing service is ceased before the port completes, the number is usually lost permanently and cannot be recovered. If the porting request details do not match the losing provider’s account records exactly, the request is rejected and the timeline restarts. And if your numbers sit on a reseller’s sub-allocation from a wholesale range holder, an additional party is involved in the process. Confirm porting-out cooperation, charges and timescales in the contract before you sign, not when you want to leave.

Does hosted VoIP work over a standard business broadband connection?

Usually, provided the connection has sufficient upload capacity and the network is configured for it. A concurrent voice call consumes roughly 85 to 100 kbps in each direction including overhead, so ten simultaneous calls need about one megabit of reliable upload with headroom for everything else the business is doing. What matters more than raw speed is consistency: jitter and packet loss degrade calls long before bandwidth becomes the constraint. Disable SIP ALG on the firewall, configure traffic prioritisation, use wired connections for desk handsets rather than wireless, and consider a secondary connection for failover if the phones are business-critical.

How do we integrate a hosted phone system with Microsoft Teams?

There are three routes and they differ materially. Direct Routing connects a provider’s SIP service to Teams through a certified session border controller, keeps Teams as the user client, and offers the widest choice of carrier. Operator Connect places a participating carrier inside the Teams admin experience with simpler onboarding but a shorter list of available operators. The third option — running a provider’s own softphone alongside Teams — is frequently presented as integration but gives users two applications that both ring. Establish which route is being offered, what Microsoft licensing it requires, and who owns and supports the session border controller if one is involved, because the licensing is where the total cost usually changes after the decision.

What happens to lift phones, alarms and card terminals when we move to VoIP?

Each of these needs its own migration route and none of them appear in a staff headcount, which is why they are the most commonly missed part of a voice project. Lift emergency telephones carry a statutory safety obligation and typically need either a GSM or an IP-based replacement unit. Intruder alarm signalling, door entry panels, card payment terminals and franking machines all need to be identified during the inventory stage and assessed individually — some can move to an analogue terminal adapter, others need replacing outright. Survey them at the start of the project, because the lead time on replacement units is frequently longer than the lead time on the phone system itself.

How do we protect against VoIP toll fraud?

Toll fraud usually follows a compromised extension password or an exposed SIP port, and the resulting calls to premium-rate or international destinations can accumulate a five-figure bill in a single night. The defences are straightforward and should be configured on day one: bar international and premium-rate destinations by default and enable only what the business genuinely calls, set a hard spend cap where the platform supports one, enable anomaly alerting on unusual call patterns and out-of-hours activity, enforce strong unique passwords on every extension, and restrict registration by IP address where practical. Confirm the provider’s liability position in the contract as well, because most agreements place the cost of fraudulent traffic with the customer.

Should we sign a three-year or a five-year VoIP contract?

Three years is the better default for most UK organisations. The per-seat discount attached to a five-year term is usually one to three pounds, which on forty users is somewhere between £2,400 and £7,200 across the term — real money, but set against committing to a 2026 platform until 2031 in a market that has changed materially in every preceding five-year window. The shorter term also limits the damage from a poor selection, which matters because support quality and integration fit are the factors most likely to disappoint and the hardest to assess in advance. If a five-year term is unavoidable, negotiate a fixed price for the duration and a permitted seat reduction band in exchange.

What should we do before contacting any VoIP providers?

Spend a day or two building the requirement. Inventory every telephone number and which provider bills it, pull peak concurrent call volumes from three months of actual call records, list every non-telephone device using a line, calculate the current total cost including line rental, call spend, maintenance and any hardware finance, confirm when the existing contract ends and what notice applies, document how calls are routed today including out-of-hours rules, and list the business systems the phone system must connect to. Providers responding to a documented requirement quote against your business; providers responding to an approximate headcount quote against their standard package and fill the gaps with assumptions you discover during installation.

Related reading

These guides cover the decisions that sit either side of a voice procurement — the copper migration that often triggers it, the connectivity and monitoring the phones depend on, and the planning and assurance work that surrounds it.

• The PSTN Switch-Off: A UK Business Guide to VoIP Migration
• Bandwidth Planning for a Growing UK Business
• Network Monitoring: A Proactive Guide for UK Business
• Building an IT Roadmap and Technology Strategy
• Cyber Essentials Plus: What the Technical Audit Covers