Dial Tone as a Dependency: Auditing the Interfaces the PSTN Withdrawal Will Expose

The withdrawal of the public switched telephone network is usually described in the language of telecoms procurement: lines, contracts and handsets. That framing is too narrow for most UK organisations. The PSTN is an integration layer, albeit an old one. It carries alarm signalling, out-of-hours telemetry, payment authorisation, fax workflows and fallback paths that may never appear on a network diagram. When the analogue service is withdrawn, those interfaces do not simply move; they are exposed.

Treating the change as a migration of integration dependencies rather than a telecoms purchase changes the questions. The first is not which voice platform to buy, but which business processes still rely on a dial tone, and what must happen when that tone is absent. By the end of 2026 the UK's analogue network is being withdrawn, so the audit window is now an operational necessity rather than a planning exercise.

The wrong frame: procurement before dependency

A procurement-led response tends to start with the replacement service. It asks which SIP trunk, hosted voice product or analogue terminal adapter will keep the most visible lines working. That is a reasonable question, but it is not the first one. The first question is what the line actually does. A telephone number on a bill is not evidence of a voice service. It may be the return path for a lift alarm, the only out-of-hours channel from a remote pumping station, or a fallback route for card authorisation when the primary broadband circuit fails.

These are integration dependencies. They have inputs, outputs, failure modes and escalation rules, even if nobody documented them as such. The withdrawal of the PSTN is therefore a controlled change to an interface contract that many organisations never knew they had. The audit should be governed as an integration programme, with the same rigour applied to retirement or platform replacement, not delegated solely to a telecoms buyer.

Evidence before inventory: reading the line estate

The most useful starting point is not a questionnaire sent to budget holders. It is line-level evidence. Call detail records can show which numbers are active, which are silent and which follow a pattern consistent with alarm signalling or scheduled telemetry. Billing records can reveal lines that are paid for but absent from any current site plan. Engineer notes and site surveys can confirm what is physically connected, while test calls can establish routing and answer behaviour.

Evidence does not remove the need for human knowledge, but it sequences it. A line that has not been dialled for months may still be the only path for a lift alarm, because the alarm only calls when something is wrong. Conversely, a line with regular traffic may be a redundant voice channel that can be retired without operational consequence. The audit should distinguish active use from dormant dependency, and it should record both. The output is not a list of phone numbers; it is a dependency register with owners, sites, connected equipment and failure expectations.

Failure semantics: what the replacement must preserve

The central risk in an analogue-to-digital migration is not loss of voice quality. It is loss of failure semantics. An analogue line has particular behaviour when the exchange is unavailable, when the line is seized by an alarm panel, or when power fails at the premises. A digital replacement may preserve the dial tone in normal operation and behave differently when the broadband circuit drops, when the adapter loses power, or when the alarm panel attempts to seize a channel that is now shared with voice traffic.

That difference matters most in alarm routing and escalation. A fire alarm that dials a monitoring centre, a lift alarm that calls a response desk, or a telemetry unit that raises an out-of-hours fault all depend on a sequence: detection, seizure, dialling, connection, identification and escalation. If the replacement changes any step, the operational outcome can change even when the line appears to work. The specification should therefore describe the required behaviour under failure, not just the required service under normal conditions.

From audit to acceptance: specifying the digital substitute

The replacement design should be tested against the dependency register. For each critical line, the acceptance criteria should include a simulated failure: the alarm activation path, the out-of-hours routing, the escalation chain and the fallback if the primary network is unavailable. It is not enough to confirm that a call can be made. The organisation needs evidence that the same operational response is triggered when the call is made under realistic conditions.

This is where integration strategy and telecoms delivery meet. The chosen digital service may be a SIP trunk, a hosted voice platform, an analogue terminal adapter or a managed alarm solution. Each has different dependencies on power, broadband, local cabling and monitoring centre compatibility. The audit should record those dependencies explicitly, so that the replacement is not simply a like-for-like dial tone but a tested substitute for the behaviour the organisation relies on.

Governance at the edge of the estate

The lines most likely to be missed are at the edge of the estate: temporary sites, remote equipment, legacy payment terminals, fax workflows and fallback routes that nobody has needed to use for years. They are also the lines most likely to fail silently after migration, because no one is watching them in normal operation. The governance response is to assign ownership at the dependency level, not the line level. A site manager, an alarm provider, a facilities team and a network team may each hold part of the picture.

The PSTN withdrawal is not only a telecoms event. It is a forced review of old interfaces that many organisations have inherited rather than designed. An evidence-led audit, followed by acceptance testing that preserves failure semantics, turns that review into a controlled migration. The alternative is to discover the dependency at the moment the line is needed and the dial tone is no longer there.