Summary

  • Old plans with active customers would stay in the records, allowing their terms to be checked against current bills.
  • A stable customer code would support comparisons between returns; keeping names out would not settle every question about linking the data.

A broadband offer has a headline price. A bill has discounts, extras and usage charges. They describe the same purchase from different ends, and neither is much use for explaining the difference on its own. ACCC wants to bring them together.

Its September 2026 proposal would collect separate records for individual telecom services and the plans attached to them. A plan ID would connect the two. That seemingly modest link is the useful part: it could let Australia's competition regulator examine what was billed against the terms of the relevant offer, rather than treating the advertised price as the customer's whole experience.

The consultation closed on 7 October. As checked on 19 October, the official project page still describes an ongoing proposal, not an adopted rule. The September draft proposes March and September reference months, with the first September 2027 return due on 11 November 2027. Those are proposed dates; two returns a year would not mean collecting six continuous months of usage each time.

The offers that have disappeared from the shop window

The proposed coverage includes retail broadband, fixed voice and mobile services, alongside broadband and mobile plans with active customers. That last qualification matters. An old plan may no longer attract new customers but still appear on existing customers' bills. Leaving it out would leave those services without their matching terms.

For broadband, service records would include the brand, connection date, access carrier and technology, and a previous provider where known. SA1 would identify a statistical area, not a street address. Charges billed, discounts, add-ons and usage would sit beside those details. Plan records would supply speeds, charges, contract terms and availability. Billed amounts are not proof that the money was paid.

The scope also needs a careful reading. The paper uses 50,000 services in operation across technologies as a criterion for adding record keepers, but the draft names providers and includes selected smaller-market operators. It does not give every business below 50,000 an automatic exemption. Who appears in the eventual returns will shape how much of the market the comparisons describe.

Joining the records could expose differences between plans, places and access technologies. It would not explain every difference. A larger bill is not, by itself, evidence of unfair treatment; nor is a usage record a speed test. The joined data would give ACCC better questions to investigate, not ready-made answers about causes or network quality.

Remember the customer, without receiving the name

A second identifier would do a different job. The proposal calls for a scrambled customer ID that stays consistent across returns. Unlike the plan ID, it would support comparisons over time. The text establishes continuity in a provider's reports, not a common key that automatically follows someone across every operator.

The customer key must not be the provider's original customer ID, allow that ID to be recovered or personally identify the customer. Disclosure would be restricted to the record keeper and ACCC. A secret-key hash is suggested as one possible method; the paper does not present a finished, independently tested implementation.

ACCC says it did not undertake a full privacy impact assessment because the proposed records no longer require personally identifiable information. It also says its protections for commercially sensitive information would continue to apply. That is its rationale, not evidence that joining records creates no risk. The feature that makes the data useful—remembering which records belong together—also makes access controls and the protection of the key consequential.

The unresolved choice is therefore larger than which columns to collect. It is how much continuity the regulator needs, and how tightly that continuity can be contained.

Sources