Blog · Data quality and reconciliation · Telecoms and connectivity
Site penetration is only as good as the site count it divides by. This guide sets out where a business telecoms provider gets each customer's estate, how to date and source it, how to reconcile it to billing so closed sites do not linger, and the review cadence that keeps the number honest.
Site penetration, sites on-net over sites in the estate, is the number that tells a business telecoms provider where its next sales are. Everyone gets the numerator right; it is in billing. The denominator, the customer's estate, is a list somebody typed into the CRM at the last tender and nobody has touched since. This guide sets out how to source, date and maintain it.
| Source | Quality | When |
|---|---|---|
| The customer's site list from a tender | High, dated | At tender |
| The account manager's review with the customer | High, dated | At each review |
| Public register of trading addresses | Medium | When the customer will not share |
| Norm for customers of that size and sector | Low, labelled | Fallback |
Every customer's estate carries the source and the date. The penetration figure inherits both.
Two checks, every month:
billed sites ⊆ estate
The first check is an assertion; the second is a review list.
One customer, estate last refreshed twenty months ago.
| Estate as recorded | Billing this month | Reconciliation |
|---|---|---|
| 42 sites | 38 sites billed | 3 billed sites not on the estate |
| 7 estate sites with no billing for 14 months |
Penetration as reported: 38 of 42, 90 percent. After the refresh: the three unknown sites added, the seven confirmed closed by the account manager, estate of 38, all billed. Penetration 100 percent, and the seven closed sites drop out of the account manager's gap list, where they had been the top three lines for a year.
Estate never refreshed. The most common state. The days-since-refresh column is what changes it.
Closed sites kept. They make penetration look low and the gap list long, and the account manager stops trusting it.
Source not recorded. An estimate from a register treated as a customer-confirmed list. Carry the source.
Billing site identifiers that do not match estate identifiers. The reconciliation cannot run. One site identifier per address, in both systems.
Mapped once, billing and the estate produce the two reconciliation lists and the staleness ranking every month, and penetration is reported with its estate date. Covirage builds this from the exports as they are. The telecoms page describes the setup, and the site penetration guide covers the measures the estate feeds.
At every account review, and whenever billing shows a site the estate does not know. Annually is the floor. The report should show the date of each customer's last estate refresh, so a stale one is visible.
Use a public register of trading addresses or the norm for customers of that size and sector, labelled as such. The penetration figure is then an estimate, and the report says so.
Every billed site is in the estate by definition. A billed site not on the estate list means the list is incomplete; a site on the list with no billing for a year and no known competitor is probably closed. Both are exceptions the report lists.