Performance Max Audience Signals Explained

Some advertisers treat audience signals like targeting. They add a customer list, a couple of interest segments and a custom audience, then assume Performance Max will only show ads to those people. It won't.

‍

Audience signals are suggestions, not strict targeting criteria. Google states that Performance Max may show ads to relevant audiences outside your signals if they have a strong likelihood of converting. This article covers what audience signals actually do, what you can use as a signal, how to build a set that helps, and what signals can't fix.

“An audience signal is a hypothesis about who your customer is. It isn't a boundary around who sees your ads.”

What Audience Signals Actually Do in Performance Max

Audience signals are an optional input set at the asset group level. You give Google a few audiences, such as a customer list, a custom segment or a Google segment, and Performance Max uses them as a starting point to find users with similar or stronger intent. Google's developer documentation describes them this way and notes that standard Performance Max campaigns can run without them.

‍

That's the difference between a signal and targeting. Targeting is a filter: only people inside the audience see your ads. A signal is a hint: start here, then find more people like this. If the system sees a strong likelihood of conversion outside your signals, it can serve there too.

‍

Signals can be particularly useful early in a campaign, helping Google AI understand which audiences may be relevant. They can continue to inform the campaign as it runs, especially when first-party data is kept up to date.

What Can You Use as a Performance Max Audience Signal?

Performance Max supports several types of audience signals, and they provide different kinds of information about your potential customers.

Your Data

Customer Match lists, website visitors and other first-party segments. This is one of the strongest audience inputs you can give Google because it's based on your own customer and website data. First-party data can also help guide the learning phase. Without it, Performance Max has to lean on broad signals, which usually means a longer and more volatile start with higher CPA swings.

‍

One catch: a list of past buyers shows Google what your existing customers look like. If your goal is new customer acquisition, check that the campaign isn't just finding more people who already know you.

Custom Segments

Custom segments describe intent through search terms and URLs: what your ideal customer searches for and which sites they visit. They let you build a signal around your product instead of relying on Google's pre-built categories. A furniture brand, for example, might build a segment from searches like “velvet sofa” and from the URLs of competing sofa retailers.

‍

Competitor brand names, keywords and URLs are a common starting point. In one apparel eCom account, we launched a dedicated Performance Max campaign using competitor brand names, keywords and URLs as custom audience signals. It met the performance targets and stayed live after the test phase.

Google Audience Segments

Interests, in-market segments, and demographics. These are Google-defined signals, so they're broad by design rather than tailored to your business. They're useful when you don't have meaningful first-party data yet, or as a supporting signal alongside stronger audience inputs. On their own, they give Google less signal than your own data or a well-built custom segment.

‍

How to Build Audience Signals That Actually Help

A good signal set is specific and consistent with the asset group it sits in. A bad one is a long list of everything available. Use this checklist:

‍

• Separate brand and prospecting first: In a mixed campaign, branded traffic can make performance look stronger than it is. If new customer acquisition is the goal, use the appropriate brand controls and evaluate performance separately. 
• Start with first-party data where you have it: It's the strongest input you control. If you don't have a usable list yet, begin with a custom segment and add your data as it builds.
• Match signals to the asset group: An asset group is a theme: products, creative and messaging. Signals should describe the person that theme is written for. If the audience would respond to a different message, that's a reason for a separate asset group, not just another signal. Mismatched assets and signals send the system in a mixed direction.
• Don't overcomplicate the set: Every signal should answer one question: why would this person buy this product? If you can't answer it, leave it out. Google treats signals as optional, so there's no need to fill every field.

What Audience Signals Can’t Fix

They can't work as targeting. Because Performance Max can serve outside your signals, a signal won't reliably keep ads away from anyone. Don't build a set expecting a hard boundary.

‍

They can't fix broken tracking. Smart Bidding optimises towards the conversion data it receives. If purchases are duplicated or values are missing, a perfect signal set still points the system at the wrong outcome. Get conversion tracking right first.

‍

They can't replace a strong product feed. In a furniture account where we rebuilt the prospecting feed from scratch, with around 120 title rules across 30 product categories, spend grew roughly 41% while ROAS held near 3.5x. The case study's takeaway: in high-AOV, competitive categories, the feed was the main scaling lever, not the bidding strategy. Signals sit on top of that foundation, not in place of it.

‍

They can't promise better ROAS. Google calls signals optional, and standard Performance Max campaigns run without them. Be sceptical of any claim of a fixed uplift from adding signals. The only way to know what they do in your account is to test.

How to Test Performance Max Audience Signals

Test audience signals as one variable. Keep assets, products, campaign structure and bidding consistent, change only the signals, and allow enough time and conversion volume before drawing conclusions. Two weeks is a reasonable minimum observation period, but Google recommends allowing around 4–6 weeks for Performance Max experiments to produce more meaningful results. Aim for 30–50 conversions before evaluating the impact, where conversion volume allows, and consider the campaign's performance and data quality before deciding whether the change made a difference.

‍

Then judge on business metrics at the campaign level: CPA, ROAS and how many of the buyers are genuinely new. Evaluating Performance Max at the campaign level is important, because the algorithm optimises against your conversion objectives.

‍

Audience insights are useful for seeing which segments your ads resonate with, but they describe who converted, not whether your signal caused it. Treat them as a source of new hypotheses, not as proof.

The Bottom Line: Audience Signals Guide PMax, They Don’t Control It

Audience signals are worth setting up, but think of them as a starting hint for Google's AI rather than a fence around your audience. They tend to help most when they come from real customer data or a clear view of intent, and when they're matched to the asset group they sit in.

‍

Starting with a customer list and a well-built custom segment is usually more useful than adding a long list of signals you can't explain. From there, let the account show you what works. Changing one thing at a time and giving it room to collect data will teach you more than any best-practice checklist.

‍

And if results don't move, that's useful information too. It usually points to something underneath the signals, such as data quality or campaign structure, that deserves attention first. Once that foundation is solid, audience signals become a sensible refinement on top of a campaign that's already learning from good data.

‍