Google Ads remarketing: the audience lists worth building and the campaigns to run them in

Google Ads remarketing is now framed as first-party “your data” audience segments, and the mechanics changed in 2023–2024. This guide maps the current taxonomy, the real eligibility numbers, and exactly where and how to use each segment type, with cross-references to how these lists seed Demand Gen lookalikes and, if you need it, to how audience signals behave in Performance Max.

Your data audience taxonomy and what replaced remarketing

Google’s help documentation now describes remarketing audiences as “your data segments.” These are first-party audience segments you own and control, built from several sources that replaced the older “remarketing list” terminology.

The current your-data taxonomy consists of:

• Website visitors: users added within seconds of visiting a tagged page, sourced from your site tags.
• Dynamic remarketing: website-based segments that also carry product or service data for dynamic ad creatives.
• YouTube users: segments based on engagement with your YouTube channel or videos.
• Customer Match: customer list segments built from uploaded first-party data such as email, phone, or address.
• App users: segments based on users of your app.

Compatibility is broad: website visitors, YouTube users, and customer list segments can be used across Search, YouTube, Display, and Gmail. App user segments cover Search, YouTube, and Display but do not extend to Gmail. These are all considered your data segments and are also eligible for exclusions and, where supported, as seeds for downstream systems like Demand Gen lookalikes.

The foundational audience lists every account should build

A common assumption is that you wait to build remarketing until you have campaigns that need it. The more durable approach is to treat these lists as infrastructure: segments start accruing members and value as soon as they are created, even if no campaign targets them yet.

Per Temper And Forge’s practice, the build order is:

• Tag-based website segments from day one: all visitors, key product or service page visitors, cart or lead-form starters, and converters. Since website visitors are added within seconds of a tagged pageview, these lists begin filling immediately.
• Customer Match as soon as first-party data and consent allow: core customer file, high-value customers, and recent purchasers, uploaded in the documented CSV format.
• Funnel-stage segments that mirror qualification: for example, marketing-qualified leads, sales-qualified leads, or trial users, mapped to URL rules or Customer Match subsets.

These segments accrue value before campaigns need them because eligibility rules look at recent active users and membership windows. When you later launch Search, Display, YouTube, Performance Max, or Demand Gen campaigns, you already have 100+ recent users in key lists, so you can immediately use them for observation, exclusions, or as seeds without waiting weeks for data to accumulate.

Eligibility minimums, 100-user rule, durations, and refresh

Many practitioners still plan around the old requirement that remarketing lists need thousands of users. Google’s current documentation states that serving eligibility for your data segments is 100 active visitors or users within the last 30 days for Display, Search, and YouTube, and this applies to customer lists as well.

The 100-user rule applies to lists uploaded or refreshed after 1 February 2024. Older, stale lists that have not been refreshed still sit under the previous 1,000 matched-user minimum, so refreshing or rebuilding legacy lists is a practical step if you want them to serve under the newer threshold.

Membership durations are also bounded. Website visitor and search-based segments default to 30 days and can be set up to a maximum of 540 days. Customer Match lists have a 540-day maximum membership duration, and eligibility requires at least 100 members added or updated within the last 540 days, so Google recommends regular refreshes of customer files.

Audience lists do not last forever. Documentation specifies that unused segments close automatically after 540 days without targeting use. Combined with the 540-day membership limit, this means you should plan periodic list reviews, refreshes, and campaign-level usage so that key segments remain open, populated, and eligible to serve.

Customer Match mechanics, upload format, hashing, and access tiers

Customer Match is Google’s customer list system that reaches Search, the Shopping tab, Gmail, YouTube, and Display using uploaded first-party data. Only first-party data is allowed, data about users under 13 is prohibited, and lists cannot be defined so narrowly that they become overly specific.

Uploads use CSV files with documented headers for each identifier type. When you use mailing address as a matching key, Google’s documentation requires First Name, Last Name, Country, and Zip together for that record to be eligible for address-based matching.

Before matching, Google normalizes and hashes identifiers. Normalization includes trimming whitespace, lowercasing, formatting phone numbers in E.164, and treating gmail.com and googlemail.com equivalently by ignoring periods in the local part. Hashing uses SHA-256, and you can either pre-hash your data or allow Google to hash it on upload after normalization. Country and zip fields are not hashed.

Customer Match access is tiered. All policy-compliant advertisers can use Customer Match lists for Observation and Exclusions. To use them for Targeting and manual bid adjustments, an account must have at least 90 days of Google Ads history and more than 50,000 USD in lifetime spend, and must maintain good policy-compliance and payment history. A common misunderstanding is that anyone can target Customer Match lists from day one; the access tier rules prevent that until the account matures.

Where each audience type can be used across campaigns

Your data segments have several usage surfaces: targeting, observation, exclusions, Performance Max audience signals, and Demand Gen lookalike seeds. Website visitors, YouTube users, and customer lists can be applied across Search, YouTube, Display, and Gmail, while app users can be applied on Search, YouTube, and Display.

On Search, you choose between Targeting and Observation at the ad group or campaign level. Targeting restricts reach so that only users in the selected segments are eligible to see ads. Observation does not change who sees ads; instead, it lets you segment performance by audience and, under Smart Bidding, feeds those first-party segments as signals into the bidding system. Google explicitly recommends Observation for all Search campaigns, which is the modern form of remarketing lists for search ads (RLSA).

A frequent confusion is that simply adding an audience to a Search campaign will restrict it. This only happens if you set the audience to Targeting. In Observation mode, the campaign’s reach is unchanged, and the first-party data is used to inform bidding and reporting.

Exclusions are available for your data segments and custom segments. You can exclude website visitors, YouTube engagers, customer lists, and app users from specific campaigns or ad groups. This is central to patterns like keeping existing customers out of acquisition math where the goal is new customers, while still allowing them in dedicated upsell or retention campaigns.

Performance Max uses audiences differently. Documentation states that audiences in Performance Max are signals only, which means your selections guide the system but do not hard-limit where it can serve. These same your-data segments also act as seeds for Demand Gen lookalike segments, which are configured in Demand Gen campaigns and built from customer lists, website or app visitors, and YouTube engagers. For broader detail on how those lookalikes are targeted in practice, see how audiences fit into account structure.

What replaced similar audiences and how Demand Gen lookalikes fit

Similar audiences no longer exist in Google Ads. Google stopped generating new similar audiences in May 2023 and removed them from campaigns in August 2023. Any strategy that still assumes similar audiences are available needs to be updated to the current replacements.

Google’s documented replacements are:

• Optimized targeting for Display and Video, which expands reach beyond your manually selected segments.
• Audience expansion for certain Video reach formats.
• Smart Bidding with first-party data on Search and Shopping, where your data segments in Observation mode act as signals.
• Lookalike segments in Demand Gen, which are explicitly described as a replacement for similar audiences on that surface.

Demand Gen lookalike segments are built from a first-party seed list drawn from the same your-data sources described earlier: customer lists, website and app visitors, and YouTube engagers. You choose a reach setting such as narrow at about 2.5%, balanced at about 5%, or broad at about 10% of the available population. Google’s documentation notes that lookalikes are transitioning to suggestion mode through 2026, but they remain the lookalike mechanism tied to your first-party seeds in Demand Gen.

Consent requirements for EEA traffic and Customer Match

For European Economic Area traffic, Google’s documentation sets explicit consent requirements for Customer Match. Since March 2024, any Customer Match list used in the EEA must be based on users for whom both consent fields, ad_user_data and ad_personalization, are set to GRANTED.

This means your upstream consent capture and data pipelines must distinguish users who have granted both types of consent from those who have not, and only the former can be included in Customer Match segments that will be used in the EEA. For a broader treatment of how consent mode and conversion tracking interact with these requirements, see the consent and conversion tracking reference.

Temper And Forge treats audience lists as infrastructure that gets built before it is needed, following the same tracking-before-spend posture used for measurement. Lists accrue members and value while campaigns are still proving themselves, so the build order is tag-based website segments from day one, Customer Match as soon as the first-party data and consent footing allow, and funnel-stage segments that mirror how the business qualifies people. In operation, Temper And Forge runs observation-first on Search so Smart Bidding gets the signal without reach restriction, uses exclusions to keep existing customers out of acquisition math where the goal is new customers, and uses these lists as seeds for Demand Gen lookalikes where the social-style surface earns a test, with ROAS still deciding everything while audiences sharpen the signal rather than replace the read.