A Coverage Number You Cannot Prove: The Records Problem Inside Canada's Broadband Push

Canada has committed, in federal policy, to 98% household broadband coverage at 50/10 Mbps by 2026 and universal coverage by 2030. The $3.225-billion Universal Broadband Fund is the primary vehicle, and provincial co-funding programs are the delivery arm. In principle this is an engineering problem: trench, splice, light the fibre, move to the next address. In practice it is a records problem first, and the engineering is how you produce the evidence.

Federal funders do not pay for trenches. They pay for connected, serviceable households, and they pay only when the claim is substantiated. Between those two facts — build the infrastructure, prove the connection — lies the gap where broadband programs routinely lose time, money, and credibility.

The Size of the Remaining Problem

The gap between urban and rural connectivity in Canada is not a rounding error. Approximately 99% of urban households have access to 50/10 service. In rural areas the figure is roughly 46%; on First Nations reserves it is approximately 35%. The addresses that make up the final reach of a broadband program — the ones that move a coverage map from 96% to 98% — are precisely the ones where the terrain is hardest, the contractor oversight is most remote, and the records are most likely to be incomplete.

These are also the sites most likely to be sampled in a federal audit. A program that can accurately report 50,000 connected households in aggregate but cannot produce the eligibility determination, the completion certificate, and the serviceable-address count for any given property on demand is a program with a coverage figure it believes but cannot defend.

Coverage Is a Claim Before It Is a Connection

Every site in a provincial broadband program moves through the same evidence chain: the address is assessed as eligible for funding, a funding agreement is signed, construction milestones are reached and certified, a completion certificate is issued, a serviceable-address count is verified, and a coverage polygon is drawn that reflects what was actually built. Each of these steps generates a document. Each document has to be tied to the specific site it describes, current as of the day it was generated, and retrievable on demand.

Where programs fail is in the gap between a contractor's progress update — "site complete" — and the full substantiation the funder requires. A site marked complete in a spreadsheet is not the same as a site with a signed completion certificate, a serviceable address count reconciled to the agreement, and a coverage polygon that matches field conditions. When those three things do not reconcile, a subsidy claim stalls and clawback exposure rises.

The Multi-Party Accountability Problem

A provincial broadband program does not build anything itself. It funds internet service providers who build, and it co-funds alongside federal programs that each have their own reporting requirements. This creates a layered accountability problem: the province is responsible for program delivery, but the evidence of that delivery sits primarily in systems controlled by ISPs — contractors who have no particular incentive to structure their records to satisfy federal audit requirements.

A GIS layer that is updated from ISP progress reports rather than from site-level completion evidence will lag field reality by weeks or months. A program director who manages thousands of sites across dozens of ISPs and relies on email chains for progress updates is not managing a program record. They are managing a correspondence archive and hoping the two things match.

The 98% Target Is a Records Commitment

Innovation, Science and Economic Development Canada's published UBF target — 98% by 2026 — is a coverage commitment that will be verified at the address level. Programs that hit the number will be the ones whose site-level evidence is organized so that any sampled address can be substantiated immediately: eligible, funded, built, connected, serviceable. Programs that report 98% from aggregate data but cannot decompose it to the site level are reporting a belief, not a proof.

The last two percent of a coverage map — the rural and remote households that everyone agrees should be connected — are the hardest to build and the most contested when audited. They are also the only ones still outstanding. Every governance decision about records that a program defers to next quarter compounds against that final stretch.

What a Site-Level Records Discipline Requires

• Make the address the unit of record. A coverage number that cannot be decomposed into substantiated individual sites is not auditable; every claim starts at the property level.

• Tie the subsidy claim to its evidence at the time of the claim — completion certificate, serviceable-address count, coverage polygon — not retroactively when an audit requests it.

• Reconcile the GIS map to field conditions, not to ISP progress updates. A polygon that represents what was designed rather than what was built overstates coverage quietly and creates clawback exposure later.

• Control the evidence inside the program's own record. ISPs deliver the infrastructure; the program is accountable for delivery. Capturing ISP completion evidence within the program record — not only within the provider's system — is the difference between accountability and exposure.

• Assume every site will be sampled. Co-funded programs attract federal scrutiny; the record discipline required to answer a random sample with confidence is the same discipline that protects every draw in the pipeline.

The Universal Broadband Fund's 2026 target is close enough that the programs still working toward it are the ones covering the hardest terrain with the least margin for error. A coverage claim built on a site-level record — where eligibility, construction, and connection status are current and reconcilable for every address — is the only one that will hold up when the national tally is taken. You cannot count a connection you cannot prove.

Source: XNM Technologies — "The Last Two Percent: Why Rural Broadband Lives or Dies on Its Record" — xnm.ca

This content was generated by AI.