Legitimate interests assessment: planning-register data

Published alongside our privacy notice, referenced there under "Our assessment". Draft, awaiting solicitor sign-off. Companion document: data protection impact assessment.

Processing assessed: collecting, classifying and redistributing UK local-authority planning application records (address, description, decision, dates, council reference) to paying scaffolding-trade subscribers as a lead-generation digest.

1. Purpose test - is there a legitimate interest?

ScaffSignal's interest: running a paid B2B product that tells scaffolding firms which newly approved planning applications in their area are likely to need scaffold, so they can approach the site before a competitor does. This is a real, ongoing commercial interest, not speculative. Third-party interest: the subscriber firms have the same interest - earlier, more targeted sales leads than manually trawling the public register.

2. Necessity test - is this processing necessary for that purpose?

  • The register data is the only source of "an application was approved, here, for this kind of work" - there is no way to deliver the product without processing site addresses and descriptions. A less intrusive method (e.g. only aggregate counts) would not let a subscriber actually contact the site, which is the product.
  • We already minimise: applicant names and applicant contact details are stripped on ingest; only what is needed to identify the job and the council file is kept (address, postcode, description, type, decision, dates, council reference/link).
  • We do not collect more than the council itself already publishes on the public register for that application.

3. Balancing test - do the individual's interests override ours?

This is the part that needs more than "it's public and has no applicant name." Working through it honestly:

What would a resident reasonably expect? A resident who submits a planning application expects: (a) the application to be a public record, open to inspection, and (b) some degree of public interest - neighbours objecting, local press, other trades scoping nearby work. They would plausibly not expect their decided application to be picked up by an automated service and sent, same-day, to every subscribing scaffolding firm covering their council as a sales lead. That is a reasonable, not a hypothetical, privacy expectation gap, and we treat it as real rather than argue it away.

Scale. Per subscribed council, the approved-job list would go to however many firms subscribe to that council. As of this assessment there are zero live subscribers at all - the service is pre-launch/pre-revenue (see section 6 for the review trigger once that changes). When it is sent, it is sent once, as part of the ongoing daily digest for that council, not re-sent or resold, and is not a one-to-many broadcast beyond the subscriber base for that specific council.

Repetition. A given application appears in a digest once (new approval) and, since the "every job delivered" catch-up fix, may appear a second time only if it reached us late and is being delivered on catch-up - never repeated beyond that. It is not re-surfaced week after week. Cold-outreach emails about the product (distinct from the digest itself) are capped at the firm level and are limited to verified corporate bodies, never sole traders or partnerships - that protects the recipient firm, not the applicant, and is noted here for completeness.

Sensitive content in descriptions. Council-written free-text descriptions occasionally reference a named individual (rare) or incidentally disclose something sensitive (e.g. a reference to a disability-access alteration or a listed-building hardship). We now run an automated screen (section 5) that catches titled names and health/disability-adjacent terms tied to a personal reference, reducing but not eliminating this risk - an untitled name or an unrecognised phrasing can still get through. Mitigated, with the residual risk recorded and reviewed on a set date (section 5). This paragraph is about the Article 6 balancing test only. Whether an Article 9 condition is needed, and available, for the health/disability content that does sometimes appear is a separate, still-open legal question - not something this screen or this LIA resolves; see the DPIA, section 5.

Objection and control. A resident can ask to have their property suppressed from future digests and can object to the processing on legitimate-interests grounds under UK GDPR Art. 21. This is a real, working control (a suppression list), not a theoretical right - but it is reactive (the resident has to find out and ask), which weakens how much weight it can carry in the balancing test on its own.

Conclusion. The low-intensity, redaction-practised, single-council, once-per-application nature of the processing weighs in our favour; the lack of any proactive notice to the resident weighs against, as does the residual, now-screened-but-not-eliminated free-text risk. On balance, legitimate interests supports this processing, with the mitigations in section 5 built and the remaining residual risk explicitly recorded and owned. Review date: 14 Dec 2026, matching the DPIA's review date (not a separate date) - see that document, section 5, for why that date and not a longer one.

4. Why legitimate interests rather than consent

Consent is not practical here: there is no relationship with the applicant to ask them, and the processing is of a public record we have a right to access and a genuine interest in redistributing in classified form. Legitimate interests, not consent, is the correct basis subject to the test above.

5. Mitigations - actions arising from this assessment

  1. Keep the applicant-identity stripping already in place; verify it remains correct on any ingest-schema change.
  2. Built, 9 Nov 2026: an automated free-text screen that replaces titled names, and health/disability-adjacent terms tied to a personal reference, with a marker before the description reaches a subscriber's inbox or the AI classifier. Pattern-based, not a certified detector - an untitled name or an unrecognised phrasing can still slip through. The residual risk is recorded, owned by Convert and Profit Ltd, and due for review by 14 Dec 2026 - see the DPIA, section 5, for the full incident assessment and current status.
  3. Keep the resident suppression route working and reachable (privacy notice, section 4; already live).
  4. Re-review this assessment if: subscriber count per council grows materially, the product starts sharing description text with any third party beyond the Anthropic classification call already disclosed, descriptions are found in practice to carry sensitive content more often than "rare," or a real instance of the screen missing a disclosure is found.

6. Review trigger

Re-assess this LIA when: (a) cold outreach is first turned on at scale, (b) subscriber count passes 50 live subscriptions, or (c) twelve months from the solicitor's sign-off, whichever is first.