Broadband coverage: test serviceable locations before take-up
A fiber route can pass a building without making every location in that building commercially serviceable. Construction completion, building access, installation readiness and an active subscription are separate milestones. A penetration claim becomes difficult to interpret when the numerator counts active connections while the denominator counts a broader, differently defined construction footprint.
The research deliverable is a location-level availability funnel and a construction-cohort take-up table. It should identify which locations have a supported offer of service, which remain blocked, and how long each comparable cohort has been open for orders. It should not assume that a provider’s coverage report independently verifies every installation experience.
Read the availability definition, then preserve it
The FCC Broadband Data Collection FAQs explain that fixed availability reporting uses broadband-serviceable location identifiers from the Fabric, distinguishes availability from subscription data, and considers whether a location can receive a standard installation. Locate the questions on active versus inactive service and on restrictions imposed by a building owner. These are U.S. FCC reporting concepts, not interchangeable definitions of every issuer’s homes passed metric.
Start with the company’s published definition of passed, ready for service, marketable, connected and subscribed. Copy the operational meaning into the evidence register. Then inspect whether the same unit appears throughout: buildings, dwellings, business locations, accounts and connections can differ. A building with twenty units is not necessarily one serviceable customer location.
Use location identifiers where authorized data is available. When only aggregate filings exist, preserve the aggregate limit and build a hypothetical location funnel separately. Do not claim a match to the FCC dataset merely because the issuer and the map show similar geographic totals.
Reproduce the serviceability funnel
Assume a fictional network reports 10,000 passed locations. Of these, 800 are blocked by building-access restrictions and 1,200 are not yet ready for a standard installation. For this case, the two groups are mutually exclusive, the location count is accurate, and all remaining locations can receive the assumed service. Active subscriptions total 2,400 on the same location basis.
| Stage | Locations | Interpretation |
|---|---|---|
| Passed footprint | 10,000 | Construction-related claim |
| Building-access restriction | −800 | No supported installation access |
| Installation work incomplete | −1,200 | Not ready under the assumed standard |
| Serviceable locations | 8,000 | Compatible take-up denominator |
| Active subscriptions | 2,400 | Connected numerator on that basis |
Take-up among serviceable locations is 2,400 / 8,000 = 30%. Dividing by all passed locations gives 24%, which answers a broader footprint question. Neither ratio should replace the other without explanation. The first assesses conversion where the assumed offer can be fulfilled; the second also reflects the unfinished or inaccessible footprint.
For each excluded group, record the required next milestone rather than simply deleting it. Building access might need documented permission; installation readiness might need construction, equipment or a completed connection path. These exclusions affect operating timing, so the analyst should retain their quantity, expected timing where supported, and unresolved dependency.
Do not subtract overlapping exclusions twice. If 300 of the inaccessible locations are also among the 1,200 unfinished ones, the unique exclusion is 800 + 1,200 − 300 = 1,700. The serviceable remainder becomes 8,300 under those revised assumptions. An attractive funnel can be wrong because its stages were not defined as either mutually exclusive categories or nested filters.
Separate maturity from construction mix
Now divide the 8,000 serviceable locations into an older cohort of 5,000, open for eighteen months, and a newer cohort of 3,000, open for three months. Assume 2,000 old-cohort and 400 new-cohort subscriptions. Take-up is 40% and 13.33%, respectively, and the combined rate remains 30%.
Suppose the provider next adds 2,000 newly serviceable locations with no subscriptions yet. Existing cohorts retain exactly the same customers. The group take-up rate falls to 2,400 / 10,000 = 24%. No existing customer departed in this example. The fall comes entirely from adding fresh eligible locations before they have had time to convert.
That observation should trigger a maturity question rather than an immediate churn conclusion. Compare the same build cohort at similar months since service readiness. Keep its opening date distinct from construction start, announcement or first invoice. A location cannot accumulate meaningful selling exposure while a required access agreement still blocks installation.
For an actual issuer, the public data might not support cohort-level conversion. Report that limitation. Do not manufacture cohort dates by dividing aggregate capital expenditure by average construction cost, because a cost assumption cannot establish when a specific customer could order service.
Counterexample: one building, several accounts
Consider a building with twenty dwellings. The provider has access to the building, but an institutional agreement covers ten units under one billed account while six other units subscribe individually. On a dwelling basis, sixteen of twenty units have service, or 80%. On an account basis there are seven accounts. Seven divided by twenty gives 35%, but it mixes billing objects with dwelling locations.
Conversely, one business location might have two connections for redundancy. Two connections do not create two new serviceable locations. The correct comparison depends on the question: location adoption, account acquisition and circuit sales are different economic tasks. Preserve the raw measures and add a supported mapping rather than relabeling them all customers.
A separate boundary is speed and technology. Availability at one advertised service tier does not establish availability at a higher tier or through the same technology. Keep technology and relevant offered speed beside the location count, especially when a network upgrades existing locations without extending geographic coverage.
Verify the operating claim with concrete evidence
| Layer | Evidence | Action |
|---|---|---|
| Location identity | Defined location IDs or disclosed counting method | Remove only supported duplicates |
| Passed construction | Company construction and coverage records | Preserve footprint definition |
| Access and readiness | Access rights and installability evidence | Retain blocked and pending categories |
| Service offer | Technology and offered tier | Match the denominator to the product |
| Adoption | Active subscription basis | Reconcile units, accounts and connections |
| Maturity | Ready-for-service dates | Compare compatible cohort exposure |
If a company map and an FCC availability record conflict, check the observation dates, technology, location version and standard-installation meaning first. A newer map can reflect later work; a difference can also remain unresolved. Neither a missing match nor a failed individual installation establishes that an entire reported footprint is false.
Record whether installation evidence is provider-reported, independently challenged, completed for a sampled location or merely an advertised offer. These roles support different conclusions. Keep a known installation failure in its actual location and date rather than generalizing it across all locations or silently removing it to improve conversion.
Set update conditions before the next quarter
Revisit the funnel when access restrictions are resolved, construction becomes ready, location identifiers change, the company revises its homes-passed definition or a new subscription metric appears. For each movement, identify whether the denominator changed, the numerator changed, or a counting basis changed.
An evidence-ready conclusion might state that serviceable take-up improved within the older cohort while group take-up fell because new ready locations expanded the denominator. That conclusion remains conditional on the disclosed location and subscription definitions. It is more informative than equating every passed location with an immediately sellable connection or treating a group percentage as a direct churn measure.
Audit location-version changes before counting expansion
Suppose an updated location file replaces five hundred old identifiers with six hundred new identifiers after a multi-unit structure is corrected. The one-hundred increase is not automatically construction expansion. Match the old and new physical locations and classify additions as newly built, newly recognized, split, merged or unresolved.
Preserve a crosswalk rather than deleting the old version. Subscriptions can remain economically unchanged while the official location denominator is revised. Recompute both the reported-current rate and a comparable-cohort rate where the mapping supports it. If no reliable crosswalk exists, identify the break in the series instead of claiming a precise organic coverage growth number. This separates improving measurement of the footprint from actually making additional premises serviceable.
Check whether a reported subscriber can hold multiple services at one location or whether a bulk account covers several units. Preserve the mapping through cancellations and upgrades. A product cancellation should not be counted as a newly unserviceable address unless there is separate evidence that the installation offer itself ceased.
FAQ
Does homes passed mean an installation can be completed now?
Not under every issuer’s definition. Access and standard-installation readiness must be checked.
Can take-up fall without any customer leaving?
Yes. Adding newly serviceable locations can expand the denominator before subscriptions arrive.