Summary

  • In Q2 2026, Ofcom reported £874.2 million in fixed voice service revenue, 20.155 million fixed lines and 3.018 billion fixed-originated call minutes.
  • Those measures cover different things. The line total combines PSTN, ISDN and managed VoIP, while the current tables do not show the provider split that appeared for 2024.

The number that looks most like a count of “the old network” is not one. Ofcom’s headline says UK fixed lines fell by 522,000, or 2.5%, in Q2 to 20.2 million. The report explicitly says that total includes public switched telephone network (PSTN) lines, ISDN channels and managed voice over IP connections. It is a count across technologies, not a copper-line tally or a measure of how many PSTN customers remain.

The accompanying figures point in the same broad direction but are not interchangeable. Fixed voice service revenue was £874.2 million, down 2.0% from the prior quarter and 16.1% year on year. Fixed-originated calls fell 16.3% year on year to 3.018 billion minutes. Fixed broadband lines, by contrast, rose 1.0% year on year to 29.474 million. The revenue, line stock, call-minute flow and broadband count each describe a different population or period. Their coexistence does not show that one service replaced another, or that the number of lines explains the revenue change.

The tables themselves reveal a second boundary. The Q2 report shows 2024 fixed revenue, exchange lines and call minutes broken out across all operators, BT, Virgin Media and other providers, alongside BT’s share. The later quarters show totals without those operator values. Ofcom’s 2025 change notice says it removed those provider data and BT’s market share from fixed telecom tables for quarters from Q1 2025. The published release therefore no longer lets a reader attribute the recent aggregate movement to a named provider.

This is not evidence that Ofcom has no operator data internally, nor does the notice explain why the public series changed. It is a limit on what this release lets an outside reader verify. An analyst cannot identify whether the decrease is concentrated in one provider, spread across the market, or shaped by a change in the mix of managed VoIP, ISDN and PSTN lines. Nor can an aggregate count reveal whether a customer still has a voice service delivered over a different access technology.

Revenue needs the same care. The fixed table labels the measure network access and call revenue, excludes VAT, and says it is not intended to include internet access subscriptions, though some element may remain. Dividing this amount by a line count would produce a ratio, not a measured price per line: the numerator may contain access revenue and some internet element, while the denominator combines different technologies and is measured at quarter end. It would say nothing reliable about a carrier’s margin, a household’s bill, or the cost of keeping an individual line in service.

For operators, suppliers and investors, the useful signal is contraction with unresolved composition. The public record can support a statement about the direction of the reported series; it cannot establish the pace of PSTN retirement, customer switching, service quality, or which provider is gaining or losing. A better public bridge would keep the technology definition, provider coverage, revision history and reporting period beside each measure. Without that bridge, the headline remains a market total—not a map of the network underneath it.

Heng Lu’s Note 36 describes BTW’s role as making systems legible through evidence rather than advocacy. That is an editorial lens only; it supplies no evidence about Ofcom’s data or UK fixed voice.

Sources