Fixed-Line Telecommunications Cost Optimisation · Technical White Paper
Fixed-Line Telecommunications Cost Optimisation
A service-ID level audit method for legacy fixed-line estates — normalising usage across uneven billing observations, testing dormancy properly, and preventing one line from being counted in two savings levers.
3 · Estate reconstruction
Six steps before any saving is modelled
Extract
Every active service record from the provider data, preserving the account hierarchy and the product or service identifier.
Classify by function
Voice line, junction line, access service, PBX component, other recurring service, or non-billable administrative record.
Map to site
Using the provider account structure together with internal branch records — because the two rarely agree.
Separate recurring from usage
So a line with zero calls is not mistakenly treated as zero-cost. This is the single most common error in the category.
Reconcile zero-spend records
Four PBX-related identifiers here carried no spend. They required classification but created no direct cancellation saving.
Build an exception list
For services whose location, purpose, overlap or continuing requirement cannot be established from the bill alone.
4.1 · Normalisation
The equations, and the controls on them
Scroll table sideways →
| Measure | Formula or control |
|---|---|
| Call minutes | Call seconds ÷ 60, per call category. |
| Effective rate per minute | Usage charge ÷ call minutes, subject to zero-usage controls. |
| Normalised monthly usage | Sum of valid observed usage charges ÷ number of valid observations for that service. |
| Annual recurring cost | Monthly recurring charge × 12. |
| Annualised usage cost | Normalised monthly usage × 12. |
| Annual combined service cost | Annual recurring + annualised usage. |
Normalisation is necessary because service records did not all have the same number of valid billing observations. Raw totals would overstate services with more observations and understate those with fewer, so per-service averages are taken before annualisation.
The rule that stops double counting
Where a service is to be cancelled, subsequent tariff optimisation is no longer a valid independent saving lever, because the service will no longer generate usage.
The same control appears in the overlap method: remove the cancelled service from every other saving model. Without it, the same line is counted once for cancellation and again for rate alignment — which is exactly why the headline in this engagement is published as gross.
5 · Dormancy
Recurring cost exists where usage does not
The no-use classification found recurring-charge services with no supportable continuing operational usage — 31.3% of the stated annual baseline. Dormant fixed-line services are a classic source of hidden run-rate cost precisely because zero usage produces no alert anywhere.
The signed cancellation schedule listed 16 service records against 37 active records in the provider estate: 43.2% of the active-record population.
That does not mean 43.2% of operational telephony capacity was removed. Service records have different functions, and some are technical components rather than standalone user lines. It shows the scale of estate rationalisation identified, not a reduction in the organisation's ability to make calls.
5.1 · Before cancelling anything
Four tests on a zero-usage line
- Confirm the service record is still active in the provider estate.
- Confirm there is no valid operational dependency on the service.
- Check whether zero usage reflects genuine dormancy — or a service used only for inbound calls, alarms, fax, emergency access or failover.
- Check whether the service is bundled with another product whose price or functionality changes on cancellation.
6 · Overlap analysis
Finding the service nobody decommissioned
Identify everything at the site
All communications services, including those supplied under other contracts — which is where overlap usually hides.
Map technical function
Voice access, broadband, PBX access, failover, alarm or fax, inbound-number dependency, data path.
Test simultaneity
Determine whether both services are genuinely required at once, or whether the older one survived only because decommissioning was never completed.
Model the full effect
Recurring and usage, including any exit charge or residual service that must be retained.
Remove it from every other model
So the cancelled service cannot contribute to a second saving lever.
What this method produces
A register that can be audited, a dormancy classification that survives challenge, a usage model normalised across uneven observations, and a set of cancellations supported by signed authorisation.
What it does not produce, without the post-change billing evidence, is a realised saving. The method and the number are at different stages of completion, and the publication says which is which.