How the Connectivity Pricing Index Works
Sources, suppression rules, framing and versioning for every statistic in the index — written so a citation can be checked, not just believed.
The publication contract
Where the data comes from
The corpus is AMVIA’s own trading history: site-deduplicated tracker records, priced email threads, provisioning records and billing truth, accumulated across a decade of UK business connectivity trading. Each published feed states its basis verbatim — for example the ECC feed’s quoted-amount statistics are built from “site-deduped tracker records + priced email threads, pooled” and the contract-status snapshot is anchored on “billed_from_anchor”.
Privacy and confidentiality
No customer identity, no supplier buy price and no row-level record appears in any published artifact — enforced in the publishing code, not by editorial care. Every cell aggregates at least 5 underlying entities; cells that cannot meet that threshold are suppressed outright or rolled up to a wider grouping, and rolled-up cells carry a note saying exactly that.
Quoted vs transacted
Price cells state their view. The current set publishes the all-quotes view — amounts as quoted to real enquiries — and quoted-market figures are never blended with transacted ones. The ECC incidence figure is explicitly “a floor, not a rate”: it counts distinct sites with an ECC over ethernet provisioning orders, so the true incidence can be higher but not lower.
Immutability and versioning
Sets are dated and immutable: 2026-08-08 will always contain exactly these numbers, verifiable against the published manifest’s checksums. Corrections are published as a new dated set. Definition changes bump the method version (currently v1.0.0), so a cited figure always identifies the rules it was computed under.
Licensing
Headline feeds (ECC, delivery lag, contract status) are licensed CC BY 4.0 — reuse freely with the attribution "Source: Amvia UK Connectivity Data" linked to this index. Regional benchmark cells are published here for reference; contact AMVIA to licence reuse.
The index hub is here; the flagship excess construction charges page shows the contract in action.
Methodology questions
Because it counts distinct sites where an ECC was recorded over 1313 ethernet provisioning orders. ECCs that were quoted but not recorded against a tracked order don’t count towards it — so the published 1.6% (n=21) can understate the true incidence, never overstate it.
The cell is rolled up to a wider grouping (or suppressed), and the published cell carries a note naming the roll-up. Nothing is interpolated and no cell below n=5 is ever published at its own granularity.
As a new dated set. Published sets are immutable — the set you cite today stays byte-identical at its date, verifiable against its manifest checksums.