The Housing Clause Behind the Missing Settings
For advertisers in the United States and Canada, the audience panel that worked for retail campaigns no longer offers what you expect. Google's privacy-related publisher rules bar using Google platform products or Google ad code to select or target personalized ads — or to collect and use audience data such as cookie lists — on the basis of gender, age, parental status, marital status, or ZIP code when advertising housing, including for-sale or for-rent listings, and real estate services. Two details matter. It is a targeting prohibition, not a content prohibition: rental listings can still run; they simply cannot be personalized along those five lines. And the clause sits inside a wider privacy framework, so removing a checkbox does not remove the obligations that travel with personalized advertising generally.
Reading the Five Attributes as One Decision
Treat the list as one design decision rather than five separate limits. Gender, age, marital status and parental status describe who a renter is; ZIP code describes where a renter lives at a resolution fine enough to stand in for neighborhood demographics. The policy text states no rationale, but it places ZIP code alongside the demographic attributes rather than alongside ordinary geography — a signal for how to plan. What this changes:
- Household composition cannot drive audience logic or creative; "couples" and "families with children" are out.
- Life-stage bands built on age are unavailable, so seasonal moves matter more than milestones.
- Fine-grained location logic must be rebuilt on constructs the clause does not name. City, metro and radius are not listed, but widening geography purely to route around a ZIP restriction deserves a second look and, ideally, a legal review.
- Interest and contextual signals become the working material.
Where Rental Marketers Usually Slip
The prohibition covers selection and targeting; it does not bless a workaround built from proxy data. An audience list assembled from lease applications, resident portals, or a CRM export tagged with age or household type is still audience data used to select personalized ads, whatever the file is called. The same applies to inferred attributes: a model that predicts "has young children" is an inferred parental status, not a neutral propensity score. A second error is treating the five-attribute rule as the whole rulebook. Other provisions bar personalized ads built on sensitive information — health or medical records, negative financial information, ethnic or racial origin, religious belief, criminal behavior, political affiliation, union membership, sexual orientation — and on activity within adult, gambling, or government sites and apps. Renting brushes against several of these, so clean first-party data hygiene matters more, not less.
What Still Works
Personalized advertising remains available outside those attributes, subject to the conditions attached to interest-based ads: the rights to use audience data such as cookie lists, an in-ad notice that the ad is interest-based (the Ad Options icon is the example the policy gives), and adherence to frameworks such as the Digital Advertising Alliance's Online Behavioral Advertising principles or IAB Europe's framework. Contextual placement on rental content, broader geographic constructs, and creative that speaks to a property rather than to a person remain the backbone of a compliant build.
If You Also Publish Rental Content
The same document governs your own pages. Restricted content can still carry ad code, but it attracts fewer bids and sometimes none, so a listings site heavy with sensitive personal detail may earn less than a comparable lifestyle page. Violations can lead to suspended services or a requirement to change an implementation — so disclosures belong in place beforehand.
Documentation Checklist Before You Relaunch
- A privacy policy naming cookie, web beacon and IP-address collection and use, and stating that third parties may place and read cookies in a visitor's browser.
- A visible link to Google's "How Google uses data" page where that route is used to meet the disclosure duty.
- Explicit opt-in consent, obtained before collection, for precise location from GPS, Wi-Fi or cell signals, with purposes disclosed and encryption in transit.
- No cookies set on Google domains, no personal details passed to Google, and no attempt to reverse-engineer aggregated reporting.
- COPPA handling if any part of the site or app is child-directed, plus EU consent policy compliance for European users.
Boundaries Worth Stating Plainly
The five-attribute housing clause is framed for the US and Canada; the sensitive-category restrictions are not drawn that narrowly, so a campaign that is compliant stateside is not automatically compliant elsewhere. The narrow exception for certain pre-identified US government advertisers relying on a genuine occupational qualification belongs to employment advertising, not housing, so it is not a route for rental listings. And platform policy is not housing law: fair-housing obligations come from another source and vary by jurisdiction, so confirm your audience logic with qualified counsel rather than treating a policy page as the compliance standard. Policy text also moves — Google is consolidating publisher policies into a new help center, so a saved PDF is a weaker reference than the live document.
A Practical Sequence
Label every audience list by origin: first-party, licensed, modeled, or platform-built. Strike anything keyed to gender, age, parental status, marital status or ZIP code, and anything inferred from sensitive categories. Rebuild around contextual placements, metro-level geography, and property-led creative. Confirm the interest-based notice, consent capture, and privacy disclosures before publishing, and re-check the live policy pages at the start of each campaign. The goal is not a clever substitute for a banned filter: it is to make the rental offer legible where renters already read and search, and to hold the documentation that proves it.
研究の知恵