Audience Signals: Why They're Hints Google Can Override, Not a Targeting Filter

In Performance Max and Demand Gen, an "audience signal" is not a targeting filter: it is an optional hint Google's AI is explicitly allowed to ignore when it finds a better-converting user outside it. Auditing or building a campaign that assumes signals restrict delivery leads to two budget-burning mistakes: expecting a signal to exclude users it can't, and expecting a list to serve when it's below the eligibility floor. Both are checkable before launch.

Audience types: what you can build

- Remarketing ("Your Data Segments") reaches users who already interacted with the business: website/app visitors, CRM data via Customer Match, and lookalike audiences. — https://support.google.com/google-ads/answer/2497941 - In-market audiences target consumers actively researching and comparing purchases in a category — for advertisers optimizing toward likely near-term buyers. — https://support.google.com/google-ads/answer/2497941 - Affinity audiences target long-term lifestyle/passion/interest patterns inferred from search and browsing history, not purchase intent. — https://support.google.com/google-ads/answer/2497941 - Detailed Demographics target broad populations sharing a long-term life circumstance (college students, homeowners, new parents), not immediate intent. — https://support.google.com/google-ads/answer/2497941 - Life events target users around milestones (graduation, moving, marriage, new baby, new pet) tied to related purchase decisions. — https://support.google.com/google-ads/answer/2497941 - Custom segments are advertiser-defined via keywords, URLs, and apps representing target interests or purchase intent. They work with Display, Gmail, Demand Gen, and Video, but not with Performance Max, which uses audience signals instead. — https://support.google.com/google-ads/answer/9805516 - Lookalike segments find new customers similar to an existing seed list; Google's models use the seed as a signal and can reach qualified users beyond strict similarity. The former 100-user minimum seed-list size is no longer enforced. — https://support.google.com/google-ads/answer/13541369

See Google Ads Campaign Types: What Each One Is For — Picking the Right Inventory Before You Set Up or Audit an Account for which of these signal types each campaign type accepts.

Performance Max

audience signals are optional inputs that guide the AI's optimization toward the selected goal — they are not a restriction. Google states explicitly: "Performance Max may show ads to relevant audiences outside of your signals if they have a strong likelihood of converting to help you meet your performance goals." Available signal types in Performance Max: first-party data (site visitors, app users, customer lists, video viewers), custom segments, demographics, detailed demographics, life events, affinity, and in-market audiences. — https://support.google.com/google-ads/answer/14530785

Demand Gen

audience and age targeting are treated as hints, not hard filters, when "optimized targeting" is turned on — the algorithm can then serve to users outside the defined audience lists. — https://support.google.com/google-ads/answer/15594567

Practical implication for audit/setup

a signal (or a Demand Gen audience list) is a starting point for the model, not an exclusion mechanism. To actually exclude a group of users, the relevant control is an *exclusion* or policy setting, not a signal — see How to Launch Performance Max Without Cannibalizing Brand or Wasting Spend for brand-exclusion controls that do restrict delivery. Confusing "signal" with "targeting" is also the root of one Performance Max failure mode in What Does NOT Work: Performance Max Failure Modes.

Customer Match: prerequisites and consent requirements

Customer Match lets an advertiser use its own customer data to reach and re-engage audiences across Search, the Shopping tab, Gmail, YouTube, and Display. — https://support.google.com/google-ads/answer/6379332

Data source

only customer information collected in a first-party context (own websites, apps, physical stores, or direct customer sharing) may be uploaded — third-party data sources are not allowed. — https://support.google.com/adspolicy/answer/6299717

Account eligibility

the account must have a good history of policy compliance and payment before Customer Match can be used at all. — https://support.google.com/adspolicy/answer/6299717

Prohibited uploads

data for customers under 13, data collected from sites/apps directed at children under 13, and ad content implying knowledge of sensitive personal information are all forbidden. — https://support.google.com/adspolicy/answer/6299717

EU/EEA consent (as of March 2024)

Customer Match partners must supply two consent fields, ad_user_data and ad_personalization, both set to "GRANTED," for EEA user data to be processed. Data from unconsented EEA users is not processed and cannot be used for personalization. Because consent settings apply to an entire upload job, users with different consent states must be uploaded in separate jobs. — https://support.google.com/google-ads/answer/14310715

Membership duration

list memberships expire after 540 days if not refreshed. — https://support.google.com/adspolicy/answer/6299717

Minimum list sizes and why small lists don't serve

The current minimum is 100 active users, down from the previous 1,000, applying across Search, YouTube, Display, and customer lists. — https://support.google.com/google-ads/answer/7474166

Uploaded ≠ active

not every uploaded record counts. Only users signed into a Google account (Gmail, Search, YouTube, Display) at ad-serving time count as active, so a list can appear far smaller than what was uploaded — due to inactivity, opt-outs, data-quality/matching gaps, or membership past the 540-day expiry. A list under the 100-user threshold shows a "too small to serve" status; extending membership duration or merging similar segments can fix it. — https://support.google.com/google-ads/answer/7474166

Transition detail

the 100-user floor rolled out in two phases, not all at once: Search campaign Customer Lists moved from the 1,000-user minimum to 100 in May 2025; the same 100-user floor was then extended to Display and YouTube — all networks and audience types — in December 2025. — https://searchengineland.com/google-lowers-audience-size-limits-across-ads-466816

What does NOT work: signal-based automation after the cookie reliability decline

Chrome's third-party cookie phase-out remains Google's stated plan, not an abandoned one: Google's own FAQ describes an updated timeline targeting early 2025 (subject to review by the UK's Competition & Markets Authority) rather than a decision to keep cookies indefinitely — cookies still function but their reliability has dropped meaningfully as the phase-out proceeds. — https://support.google.com/google-ads/answer/14762010

What actually degrades

remarketing lists and cross-site conversion tracking see smaller audience pools and lower match rates, even though cookies were never formally removed. — https://support.google.com/google-ads/answer/14762010

What Google says is preserved

audience exclusions, brand-safety controls (content type, digital content labels, sensitive content, 3-tier exclusions), and frequency management via Privacy Sandbox APIs; billable metrics (clicks, engaged views) are unaffected; conversion tracking is transitioning to the Privacy Sandbox Attribution Reporting API. — https://support.google.com/google-ads/answer/14762010

What is comparatively resilient

campaigns already running AI-based strategies (optimized targeting, Maximize Conversions) were less hurt by the cookie reliability drop than cookie-dependent remarketing strategies; Search campaigns on high-intent keywords are described as the most privacy-resilient campaign type overall. — https://support.google.com/google-ads/answer/14762010

This is the direct auditing takeaway: if a remarketing-list-heavy setup shows declining reach with no obvious cause, the note above is where to check first — it is Google's own explanation, not a bidding-strategy problem, so switching bid strategies (see Bidding Strategies: Manual, Smart Bidding, and When Each Applies — matching the strategy to the conversion data you actually have) will not fix it.

Related

- Google Ads Campaign Types: What Each One Is For — Picking the Right Inventory Before You Set Up or Audit an Account — which audience/signal types each campaign type accepts (custom segments work in Display/Gmail/Demand Gen/Video but not Performance Max, which needs signals instead). - How to Launch Performance Max Without Cannibalizing Brand or Wasting Spend — the actual exclusion controls to use when a signal-as-hint isn't restrictive enough, e.g. to stop Performance Max from serving on brand terms. - What Does NOT Work: Performance Max Failure Modes — treating signals as hard targeting is one concrete way Performance Max underperforms; this page supplies the mechanism, that page the failure pattern. - Conversion Tracking Fundamentals — the layer every audit and every bid strategy depends on — Customer Match and Enhanced Conversions both depend on first-party data and consent signals; broken consent setup silently breaks both audience matching and conversion measurement. - What Does NOT Work: Broad Match and Smart Bidding Without Guardrails — where automation burns budget instead of saving it — like audience signals, broad match is a Google-AI-interpreted hint rather than a strict filter; the same "hint, not constraint" mental model applies and the same data-starvation risk follows if under-resourced. </content>