PMax audience signals: how to guide the AI in the right direction

Performance Max is the most ambitious campaign type Google has ever launched. The AI autonomously manages bids, placements, audiences and creative combinations with one single goal: maximising conversions within your budget. But that autonomy comes with a trade-off. Without proper guidance, the AI can spend weeks figuring out who your actual customers are. This is exactly where audience signals come in. They are not targeting in the traditional sense, but a starting point, a compass that tells the AI: "Begin your search here." In this article you will learn what audience signals are, which types deliver the most impact, how to build them strategically for both e-commerce and lead generation, and how AdBrains automates and continuously improves this process.

What are audience signals in Performance Max?

Audience signals are suggestions you provide to the Google AI when setting up a PMax campaign. They tell the algorithm which user groups are most likely to convert. Crucially, Google is not obligated to follow your signals. The algorithm uses them as a starting point for the learning phase and then autonomously expands beyond those signals if doing so yields more conversions.

This distinction makes audience signals fundamentally different from traditional audience targeting in Search or Display. You are not excluding anyone with signals; you are giving the AI a warm start. The more relevant and sizeable your signals, the faster the learning phase and the lower the CPA in the opening weeks of a campaign. Advertisers with strong first-party audience signals typically see a learning period of one to two weeks, while campaigns without signals can take four to eight weeks before the algorithm discovers conversion-ready patterns.

A PMax campaign without audience signals is like a new employee starting their first day without any briefing. They will eventually learn what works, but the learning curve costs both time and money. Strong signals hand that employee a dossier of your best customers, a list of the most frequently asked questions and twelve months of conversion history. That makes a measurable difference from day one.

The seven types of audience signals and their value

Not all audience signals are equal. There is a clear hierarchy in the steering quality each type provides. The most impactful signals are those built on your own verified conversion data. First-party data consistently outperforms third-party segments when it comes to guiding the PMax AI toward the right user profiles.

Customer Match, which uses your own CRM email list, and lists of existing converters (purchases or lead completions) sit at the top of the hierarchy. They give the AI the clearest possible picture of who has already converted and, by extension, who else might. Website remarketing audiences, particularly segmented groups such as product page visitors or abandoned cart users, follow closely. Custom segments based on active search behaviour, for example users who recently searched for terms directly related to your product, add an important layer for reaching new potential customers who have not yet visited your site.

In-market audiences and affinity audiences, which are pre-built by Google, offer a broader and less precise starting point. They can be useful as a supplemental layer but should never be the primary signal. Demographic signals are the least specific and are best used as a secondary filter when your target audience is clearly defined by age, gender or household income.

Building audience signals for e-commerce and lead generation

For an e-commerce platform like ToetsJeKennis.nl, which sells online exams and courses, the strongest audience signal structure is built in layers. The first layer is a Customer Match list of everyone who purchased an exam or course in the past twelve months. The second layer consists of remarketing audiences of product page visitors who spent more than thirty seconds on a course detail page without converting. The third layer is a custom segment based on high-intent search terms such as "online theory exam" or "food safety course online."

For a lead generation advertiser like Clima-Active.nl, which generates quote requests for air conditioning and heat pump installations, the signal strategy shifts toward purchase intent cycles. The strongest signal is a Customer Match list of previous quote requesters and converted customers, supplemented by a remarketing audience of users who visited the quote form page but did not complete it. Custom segments built around search terms like "heat pump installation quote" and "air conditioning cost estimate" capture new audiences with clear purchase intent.

The common thread in both cases is the same: the closer the signal is to verified conversion behaviour, the more effectively it guides the PMax AI from the very first day of the campaign.

How AdBrains automates and improves audience signal management

Manually building and maintaining audience signals is time-consuming and prone to error. Customer lists go stale, remarketing audiences shrink when site traffic drops, and custom segments rarely get reviewed after a campaign goes live. This is precisely where AdBrains' own AI technology creates a structural advantage.

The audience management automation system at AdBrains creates and maintains three types of audiences on a weekly basis: PROD audiences based on current converter lists, Incubator audiences based on engaged non-converters showing strong intent signals, and RLSA audiences for Search campaigns running in parallel. Every audience is refreshed weekly with the most current data, ensuring that PMax signals never become outdated or under-populated.

Server-side signal enrichment plays an equally critical role. AdBrains builds and manages a dedicated server-side Google Tag Manager infrastructure for every client. Through Enhanced Conversions, first-party data including hashed email addresses, order values and customer IDs is enriched and sent directly to the Google Ads API. This means the Customer Match lists used as audience signals are fed by higher-quality conversion signals than standard browser-based tracking can deliver. The result: the PMax AI receives richer, more accurate profile data from the moment the campaign launches.

The multi-agent verification system at AdBrains reviews every audience signal configuration change before it goes live. Four independent AI agents assess whether the proposed signals align with the conversion strategy, the campaign type and the available conversion volume. Only when all four agents reach consensus is the change implemented. This prevents the common mistake of signals that are too narrow, which constrains the AI unnecessarily, or too broad, which extends the learning phase rather than shortening it.

For e-commerce clients like E-4motion.com, which sells new electric folding bikes, this approach means PMax campaigns never have to start from scratch. The combination of enriched Customer Match data, active remarketing audiences from product page visitors and custom segments built around search terms such as "buy electric folding bike" and "compact e-bike" gives the AI a flying start. The algorithm knows from day one which user profile is most likely to convert, reducing the learning period and improving ROAS in the critical early weeks of the campaign.

Common mistakes when setting up PMax audience signals

Despite the relative simplicity of adding audience signals in Google Ads, the same mistakes appear repeatedly. The most common is adding only a single generic audience, such as "all website visitors," which gives the AI too little information to act on. A layered approach with at least three distinct signal types is always more effective.

Other frequent errors include failing to update Customer Match lists regularly, resulting in stale data that no longer reflects the current customer profile. Lists should be refreshed at minimum monthly, and ideally weekly via the Google Ads API. Uploading Customer Match without Enhanced Conversions in place is another missed opportunity: without Enhanced Conversions, match rates are lower and the signal quality suffers.

• Always use at least three signal layers: converters, intent visitors and a custom search behaviour segment.
• Set a minimum size target of 1,000 matched accounts for Customer Match lists before relying on them as a primary signal.
• Activate Enhanced Conversions to increase Customer Match match rates and improve signal quality.
• Assign different signals per asset group when your campaign covers multiple products or service categories.
• Review and refresh all audience signal sizes at least once per month.
• Avoid broad, multi-competitor custom segments as they dilute rather than strengthen the signal.

Audience signals vs. audience targeting: a clear comparison

A frequently asked question is: if Google can ignore your signals, why bother setting them at all? The answer lies in the learning speed and cost efficiency of the opening phase of any campaign. Signals are not hard targeting, but they determine the direction in which the AI first looks. That has a measurable impact in the first weeks.

The fundamental difference: traditional audience targeting builds a wall around your target group, while audience signals in PMax give the AI a compass. The AI decides whether to stay within that compass or navigate around it, depending on where conversions actually occur. For growth-oriented campaigns, that flexibility is a clear advantage, provided the signals are strong enough to ensure the AI heads in the right direction from the start.

Frequently asked questions about PMax audience signals

Are audience signals mandatory in Performance Max?

No, audience signals are technically optional in PMax. Google allows campaigns to launch without any signals. However, without signals the AI starts completely from scratch, which significantly extends the learning period and increases initial costs. In practice, strong signals are not required but are strongly recommended, especially when using Target CPA or Target ROAS Smart Bidding strategies. The faster the AI learns, the faster Smart Bidding can optimise effectively.

Can I add or change audience signals on a live PMax campaign?

Yes. You can add, modify or remove audience signals at any time on existing PMax campaigns, at the asset group level. Adding strong signals to an already-running campaign can re-stimulate the AI to optimise more accurately, though the effect is generally less pronounced than having strong signals in place from launch. For campaigns that have been running for several months and have accumulated sufficient conversion data, the AI will rely more heavily on its own learned patterns than on the provided signals.

What is the difference between audience signals and negative audiences in PMax?

Audience signals tell the AI who to prioritise as a starting point. Negative audiences, which are available at the account level or via allowlisting requests, fully exclude specific groups from all campaigns. Examples include excluding existing customers from a new customer acquisition campaign, or excluding employees via IP address. Negative audiences are hard exclusions; audience signals are soft directional guides. Both are useful but serve entirely different purposes.

How do I know whether Google is actually using my audience signals?

In Google Ads you can view the Insights report for PMax, which shows a breakdown of audience types contributing to impressions and conversions. If you see the campaign converting beyond your specified signals, it means the AI has discovered additional patterns on its own. That is not necessarily a problem. Analyse regularly whether the newly discovered audiences match your ideal customer profile, and adjust with additional signals or exclusions where needed to keep the AI aligned with your business goals.