Guide · Telecoms & Broadband

Provisioning & activation failures

When broadband is live on paper but mis-provisioned in practice: activation faults, backend configuration, engineer evidence and remedies.

A broadband service can show as activated while still being unusable or incorrectly configured. Treat provisioning as its own evidence problem rather than letting it be dismissed as a generic Wi-Fi fault.

Key points

  • Separate activation date from the date the service actually worked.
  • Record every technical diagnosis and later correction.
  • Ask whether the fault is line, ONT, router, authentication, addressing or backend provisioning.
  • Preserve engineer notes where they contradict first-line support.
  • Check delayed-start or loss-of-service compensation separately from any other remedy.

First: what type of failure is this?

“Broadband not working” can mean several different things. A line may never activate; the ONT may be provisioned against the wrong profile; authentication may fail; the router may be configured incorrectly; an add-on may not have been applied; or the customer may be behind CGNAT when a feature requires inbound reachability. Diagnose the category before arguing about blame.

The account saying “active” is not the same as usable service

Providers often have separate order, network and billing systems. One can show complete while another is wrong. If billing began before the service actually functioned, record the promised start date, the system's activation status and the first date on which the service worked normally.

Do not accept a technical shorthand that does not explain the network

A provider may say that a static IP is required for port forwarding. That can be true as a product-policy requirement, but it is not a universal networking rule. A public dynamic IPv4 address can support port forwarding; CGNAT or provider/router restrictions may block unsolicited inbound traffic. Ask whether the connection has a public address, whether CGNAT is in use, what add-on was provisioned and what changed when the issue was fixed.

Engineer attendance can be decisive evidence

Where a field engineer de-provisions and re-provisions equipment, changes the ONT profile or identifies a provider-side configuration problem, ask for the visit outcome and job notes. A later backend fix can materially undermine an earlier assertion that the customer simply needed different equipment or an optional paid add-on.

Keep support access in the chronology

If the fault lasted longer because technical support was unreachable, include call attempts, queue times and missed callbacks. This does not automatically create a fixed compensation entitlement, but it can show avoidable delay and poor complaint handling.

Remedies can stack conceptually, but do not double-count

Depending on the facts, the consumer may need a service fix, refund of wrongly charged add-ons, automatic compensation, a billing adjustment, cancellation rights or additional complaint redress. Explain which failure each requested remedy addresses. Do not ask for the same loss twice under different labels.

Evidence worth keeping

Order and activation confirmation
Router/ONT status and screenshots
IP/addressing information where relevant
Fault and callback references
Engineer job notes
Bills and add-on charges

Delayed activation and a fault after activation are not always the same event

If the service never worked on the promised start date, preserve the order and activation record. If it worked and later failed, record the first point at which service was lost. The distinction can matter to diagnosis and to any automatic-compensation calculation.

A provider may describe an order as “complete” because an order-management system reached its final status. Ask a different question: when was the communications service actually usable? If the answer is later, ask what provisioning step remained incomplete.

Router, Wi-Fi and network provisioning are different layers

“It is your Wi-Fi” is not an adequate answer to a fault that persists over Ethernet or concerns authentication, IP addressing or network-side configuration. Conversely, a weak wireless signal does not by itself prove a network provisioning failure. Test methodically.

SymptomWhat to isolateUseful evidence
No connection at allONT/modem state, authentication, line status, provisioningONT lights, router WAN state, provider fault reference
Works by cable but not Wi-FiWireless coverage/configurationEthernet speed test compared with Wi-Fi tests
Internet works but inbound services failPublic IP, CGNAT, firewall, port forwarding, provider policyWAN address, public-IP check, router rules
Feature starts working after backend changeProvider-side provisioning/profileEngineer notes and exact change made

If support tells you to buy an add-on, record the diagnosis behind the sale

If a paid static IP, replacement router or higher package is sold as the solution, ask why it is technically required. If the eventual fix proves to be an unrelated provider-side configuration error, you have a clear basis to ask for the unnecessary charge to be reviewed and refunded.

Do not assume the add-on was necessarily mis-sold simply because a different fault was later found. Preserve what was actually said: whether the add-on was described as mandatory, merely recommended, or one possible workaround.

When an engineer contradicts first-line support

Treat the engineer outcome as new evidence. Ask the provider to reconcile the earlier diagnosis with the engineer’s findings, rather than simply accepting “it is fixed now”. Useful questions include:

  • What was misconfigured or incorrectly provisioned?
  • When did that state arise?
  • Which team or system could have detected it earlier?
  • Was any paid product added because of the earlier diagnosis?
  • Did the fault affect the recorded activation date or compensation entitlement?
  • What change was made to restore service?

Automatic compensation: check the trigger, the provider and the dates

Ofcom’s automatic compensation scheme applies only to participating residential fixed broadband and landline providers and defined events. It is not a universal compensation code for every technical problem. For delayed start, complete loss of service and missed appointments, check the current Ofcom conditions and rates against the actual chronology.

If the provider is not a scheme signatory, or the fault does not meet a scheme trigger, that does not automatically end the complaint. Billing corrections, contractual remedies, discretionary redress and ADR can still be relevant depending on the facts.

Common provisioning responses, and the next question

Provider saysAsk next
“The line is active.”When did the service first pass a usable connection test, and what systems show that?
“It is your router.”What network-side checks were completed and what result rules out provisioning?
“You need a static IP.”Is that a product policy, a CGNAT workaround or a technical necessity on this connection?
“The engineer fixed it.”What did the engineer change and what was the root cause?
“Compensation is automatic, so wait.”Which scheme trigger has been recorded and from what date?

Related telecom guidance