PMax Audience Signals: How to Guide Google’s AI Toward Your Best Customers
Performance Max campaigns promise automation, but without the right guidance, that automation can burn through your budget chasing the wrong people. Audience signals are your primary tool for steering Google's machine learning toward users who actually convert—not just click.
In this guide, I'll break down exactly what audience signals are, how they work, and how to use them strategically based on what I've seen across hundreds of accounts over 17+ years in Google Ads.
If you want to explore all the posts I have written on the topic of audience signals start here: Google Ads Signals & Value Hub
Why Most Explanations of Audience Signals Are Misleading
If you have ever read Google's documentation, or watched the thousands of YouTube videos about Performance Max, you likely still have questions. Here's why.
Google explains features — not behavior. Google Help tells you what audience signals are, but not what they do. They describe signals as "guidance," but they never explain how that guidance shapes early learning, influences spend distribution, or determines the quality of your initial data. That gap is where advertisers get confused.
Advertisers are left guessing what signals actually control. Some assume signals are targeting. Others assume signals don't matter. Both interpretations are wrong. Signals matter deeply—just not in the way people expect.
Signals only make sense when you understand the incentives behind PMax. Signals exist inside a system designed to maximize volume, not profit. Once you understand that, their behavior becomes predictable—and your job becomes anchoring the machine to your actual business economics before it wanders off on its own.
This article gives you the clarity Google Help never will.
What Are PMax Audience Signals?
Performance Max audience signals are suggestions you provide to Google's algorithm about who is most likely to become a customer. They're not hard targeting rules—they're hints that help the machine learning models understand what a good prospect looks like for your business.
These signals live at the asset group level, which means each asset group in your PMax campaign can have its own distinct audience configuration. They help speed up the learning phase by giving Google a starting point, and you can change audience signals at any time without rebuilding your campaign from scratch.
Here's how it actually works: You feed Google your first-party data (past customers, website visitors), custom segments (search terms, competitor URLs), detailed demographics, and interest-based audience signals. Google then uses your signals plus its own data from across Search, YouTube, Display, Gmail, and Discover to find more users who behave like your converters—even if they weren't in your original lists.
Quick example: A local HVAC company might add audience signals including:
A customer match list of past service customers
Custom segments built around search terms like "AC repair near me" and "emergency HVAC service"
In-market audiences for home services
Google takes these inputs and expands reach to similar users across all placements, prioritizing those who look most likely to book an appointment.
From my perspective, getting Performance Max audience signals right is one of the fastest ways to cut wasted spend in new or underperforming PMax campaigns. When I audit accounts, missing or poorly configured signals are often the first thing I fix.
Signals vs. Targeting: The Real Difference
Most advertisers conflate these two things. They're fundamentally different.
Targeting is like saying, "Only show my ad to people on this exact street."
Audience signals are like saying, "Start on this street, but explore the whole neighborhood if you find more people who look like my ideal customer."
If you want strict control, use targeting. If you want Google to learn faster, use signals
If you're more of a "hear it explained" person, this video walks through why signals aren't targeting and why small businesses need high-quality signals more than anyone.
How Performance Max Uses Audience Signals Under the Hood
When you launch a PMax campaign with audience signals, you're essentially pointing Google's AI toward your most likely buyers during the critical first 2–4 weeks of learning.
Signals are not targeting — signals are trajectory. They don't tell the system who to target. They tell the system where to start. They define the initial neighborhoods the algorithm explores. They influence the first 10–30 days of learning. They shape the early data the machine uses to decide what "good" looks like.
If your signals are vague or misaligned, the machine starts in the wrong neighborhood—and the data it collects will reflect that. Once that data enters the system, the machine optimizes toward it, even if it's economically destructive.
During the learning phase, Google's machine learning models use your signals as a starting point to determine who sees your ads. But here's the key distinction: signals influence where ads show, but they don't cap or restrict delivery. If Google predicts better conversions or a cheaper CPA from users outside your specified audiences, it will absolutely go after them.
The algorithm draws on multiple inputs to build its understanding:
Recent search behavior and search queries
On-site behavior (pages visited, time on site, cart activity)
YouTube engagement patterns
App usage data
Demographic signals (age, income, location)
Your first-party data signals from customer lists and remarketing
Timing matters. After you add or adjust audience signals, expect 10–14 days before you see significant optimization changes. This is why impatient, rapid-fire edits are risky—you may reset the learning phase before the algorithm has enough data points to optimize properly.
One frustration worth acknowledging: Google does not show performance broken out by individual signal. You can't see "remarketing list drove X conversions" versus "custom segment Y drove Z conversions." This opacity is exactly why structure and testing matter—topics I'll expand on throughout this article.
Types of Audience Signals You Can Use in PMax
This section walks through the main signal types you can feed into a PMax asset group. Understanding the options helps you build a signal strategy that actually matches your business, rather than just checking boxes.
For most small to mid-sized businesses, first-party data and high-intent custom segments are the highest-impact starting points. Broader audience signals like affinity audiences work better as secondary layers once you've established a converting baseline.
First-Party Data: How They've Interacted With Your Business
First-party data includes users who have already touched your business in some way: visited your website, filled out a form, called your office, made a purchase, or exist in your CRM. This is your own data—the gold standard for training Google on what your ideal customer actually looks like.
Concrete sources to pull from:
Remarketing lists from Google Ads or GA4 (website visitors, cart abandoners)
Customer Match uploads from your CRM or email platform
Data from past Smart Shopping or Standard Shopping campaigns
Call tracking data where available
Examples of high-value lists to create:
All website visitors in the last 30 days
All website visitors in the last 180 days
Leads who haven't converted to customers yet
High-value customers from the past 12 months
Repeat purchasers
Why are these signals so powerful? They train Google on what your actual buyers look like and let the algorithm build lookalike audience patterns across Search, YouTube, Display, Discover, and Gmail. Instead of guessing, Google starts with a template of proven converters.
In my audits, missing or misconfigured first-party signals is one of the most common reasons PMax over-spends on low-intent traffic. Business owners often skip this step because uploading lists feels tedious—but it's the single highest-leverage action you can take.
Custom Segments: Search Terms, URLs, and Apps
Custom segments let you build audiences around what people are actively searching for or engaging with online. Unlike remarketing (which targets past visitors), custom segments target user groups based on their behavior across the web.
Keyword-based custom segments: Use actual high-intent search terms that historically convert or clearly signal buying intent.
"Emergency plumber Columbus" (not "how to fix a leaky faucet")
"B2B SaaS demo software" (not "what is SaaS")
"Best orthodontist for teens" (not "braces vs invisalign comparison")
The distinction matters. You want buying terms, not research terms.
URL-based custom segments: Add URLs of competitor product pages, category pages, or landing pages closely related to your offer. Skip generic homepages—Google needs specific signals about what content matters. This is a sneaky favorite of mine: you're essentially siphoning off the intent generated by your competitors' ad spend.
App-based custom segments: When relevant, add apps your ideal customer likely uses. For B2B, this might be industry tools like Salesforce or HubSpot. For consumer electronics, it might be specific shopping or comparison apps.
Demographic Targeting in Audience Signals
Demographic signals help narrow or guide Google based on age, gender, parental status, and household income. These can be valuable when your offer has clear demographic boundaries—age-specific offers, income-sensitive pricing, parental status triggers.
Remember that in Performance Max, demographics are still "soft" signals. Google can decide to show ads outside your preferred segments if it predicts strong performance there. Avoid over-restricting demographics at launch unless you have strong historical data proving certain groups don't convert. Start broader and exclude obvious poor performers once you have data.
Additional Segments: Interests, Habits, and In-Market Behavior
These are Google's predefined audiences based on interests (affinity audiences), habits, and in-market signals—users actively researching a specific product or service category.
Focus on in-market and life events segments closely aligned with purchase decisions rather than broad interest categories. "In-market for business services" beats "Business & productivity enthusiasts" every time, because in-market signals capture recent purchase intent, not general curiosity.
A word of caution: Very broad interest segments should usually be layered in only after you have strong performance from first-party and custom search segments. They can dilute your signal quality and lead Google toward users who are interested but not ready to buy.
The Cross-Hatch Method: Building Your Audience Matrix
One of the most powerful things you can do with signals is create a "unique cross-hatch" that represents your specific customer—a matrix that exists at the intersection of multiple signal layers.
When you combine In-Market audiences (like Home & Garden) + your First-Party data + a specific affinity audience (like Pet Lovers or Thrill Seekers), you're creating a unique combination that no competitor has. You're telling the machine: "My ideal customer lives at the intersection of these three things."
Inside Audience Manager, you can layer remarketing audiences, customer match lists, and detailed demographic information to build that precise intersection. The more aligned these layers are with your real customers, the faster Google identifies the pocket where your best conversions live.
People often don't realize how specific you can get. Once Google Ads starts seeing conversions, it feeds information back to you in the Insights tab. It might look like "everyone likes your ad"—but it's that original, unique combination of data you provided that allowed the machine to find that winning audience in the first place.
How to Add Audience Signals to a Performance Max Asset Group
Signals are set at the asset group level, not the campaign level. Each asset group needs its own audience signal configuration—you can't set signals once for the entire campaign.
Prerequisites Before You Add Signals
Before setting audience signals, make sure your foundation is solid. Feeding signals into a broken tracking setup just teaches Google the wrong outcomes.
You need:
Accurate conversion tracking via Google Ads or GA4
Clear campaign goals (target CPA, target ROAS, or maximize conversions with a budget cap)
Clean account structure without duplicate or conflicting campaigns
Conversion actions should be meaningful: Configure at least one or two actions that represent actual business value—lead form submissions, phone calls over 30 seconds, purchases, booked appointments. Avoid tracking micro-conversions like page views as primary goals.
In my audits, I often pause signal work until tracking is fixed. If your conversion data is inaccurate—counting spam leads as conversions, missing phone calls, or attributing to the wrong campaigns—adding audience signals just accelerates waste.
Pre-flight check:
Primary conversion action configured and tested
Google Ads and GA4 properly linked
Customer lists uploaded and processed
Remarketing lists populated (check Audience Manager)
Campaign goal clearly defined
Step-by-Step: Adding Audience Signals in the Interface
Log into Google Ads and navigate to "Campaigns" in the left sidebar.
Select your Performance Max campaign from the campaign list.
Click into the specific asset group you want to configure.
Find the "Audience signals" section. Look for "Signals" or "Audience signals" in the right-hand panel or within the asset group editing screen.
Add Your Data first. Click to add remarketing lists and customer match lists. Prioritize high-value customer lists, all converters from the past 180–540 days, and website visitors (30-day and 180-day segments).
Add custom segments. Create or select segments based on high-intent search queries and relevant URLs.
Add demographics. Set relevant demographic parameters based on your customer profile.
Add interest and in-market audiences. Layer in narrow in-market or life events segments that align with purchase decisions.
Save your changes.
Document the date in a simple tracking sheet. When you review performance in 2–3 weeks, you'll want to correlate changes with specific signal updates.
After saving, expect 1–2 weeks before the impact of new signals becomes visible in performance trends. Don't panic if you don't see immediate changes.
How Audience Signals Impact PMax Performance
Strong audience signals deliver three main advantages: faster ramp-up, more relevant traffic, and better budget efficiency.
Faster learning phase: Signals help Google skip some of the guessing at launch. Campaigns with robust signals can reach optimal performance in days versus weeks, potentially cutting early-phase costs by 15–25% through reduced irrelevant impressions.
Improved ad relevance: When your signals match your assets (headlines, images, videos), you create coherent messaging that resonates with the right audience—increasing CTR, engagement, and conversion rates.
Better budget allocation: Signals help prioritize higher-probability prospects, reducing wasted spend on completely cold, unqualified traffic.
The tradeoffs: Over-narrow or poorly chosen signals can slow learning, restrict scale, or push Google to find "lookalikes" that don't actually match your best customers. Bad signals create bad guidance.
Enhancing Targeting Accuracy Without Hard Limits
Audience signals operate as "soft guidance." They steer the algorithm, but Google can still reach outside these audiences when it sees better opportunities. This is fundamentally different from legacy manual targeting.
Judge success by outcomes, not audience matching. The right metric is overall campaign performance—CPA, ROAS, lead quality—not how neatly traffic matches your audience lists. If Google finds relevant audiences you didn't anticipate, that's the system working as intended.
The Insights tab in Google Ads can surface some of these discovered audience categories, even though per-signal breakdowns aren't exposed. Check it regularly to understand where Google is finding success.
Budget Allocation and Wasted Spend
In PMax, you have limited control over placements and search terms. Audience signals become one of the main levers to reduce waste and inform Google about who deserves your ad spend.
Strong signals (first-party data + high-intent custom segments) push more budget toward users similar to your buyers.
Weak signals (generic interests, vague keywords, untargeted demographics) lead to high impression volume but poor lead quality.
During audits, I often see PMax overspending on low-quality query themes because the audience signals were built around brand awareness terms instead of buying terms. The campaign looks busy, but the business owner is paying for tire-kickers, not customers.
The "Fake Points" Trap: How Google's NCA Bonus Distorts ROAS
This one doesn't get talked about enough. Google's New Customer Acquisition (NCA) tool can be a bit of a trick. It adds "bonus value" to new customers in your reports—which makes your ROAS look higher than it really is, but only on paper.
If you don't know this, you might think you're getting a great return while you're actually just accumulating video-game points that don't pay the bills. The inflated ROAS number is real in the reporting interface, but it doesn't reflect actual revenue hitting your bank account.
When you're evaluating PMax performance, always cross-reference your Google Ads ROAS against your actual revenue data. If the two numbers diverge significantly, NCA bonus value is likely the culprit. Adjust your target ROAS accordingly so the machine is optimizing toward real profit, not inflated metrics.
Best Practices for PMax Audience Signals (From 17+ Years in Google Ads)
The core principles:
Start with high-intent, specific audiences
Leverage first-party data aggressively
Structure asset groups around clear themes
Align creative assets with your signals
Test and iterate without constant tinkering
Data quality beats complexity every time. Fewer, better signals consistently outperform giant, messy lists.
Start With High-Intent, Specific Audiences
Begin with the narrowest, highest-intent signals you have: past converters, hot leads, and search terms clearly indicating purchase intent—not information gathering.
Avoid launching PMax with only broad custom segments like "marketing tips" or "plumbing DIY." These flood campaigns with non-buyers who are researching, not purchasing.
Leverage First-Party Data Aggressively
Upload customer lists and build remarketing segments as early as possible, even if the lists are relatively small. Something is better than nothing.
Segment by value where you can:
High-LTV customers vs. one-time buyers
Repeat purchasers vs. first-time customers
Closed-won B2B deals vs. all leads
Data quality matters. If your lists are full of unqualified leads, spam submissions, or customers who churned immediately, you're teaching Google the wrong patterns. Clean your lists first—remove bad leads, disqualified prospects, and refund customers—before using them as signals.
Structure: One Core Audience Theme Per Asset Group
Avoid dumping every audience list into a single asset group. Instead, group signals by clear themes so you can actually diagnose what's working.
Examples of themed asset groups:
High-value customers + lookalike audience signals
Cart abandoners (remarketing focus)
Competitor brand searchers (using competitor brand names in custom segments)
Local high-intent service queries
Specific product category buyers
Limit active asset groups at first. If your budget is spread across too many asset groups, none will get enough data points to optimize properly. Start with 2–4 focused groups rather than 10 diluted ones.
Align Creative Assets With Your Audience Signals
Audience signals work best when the matching creative speaks directly to that audience's needs and language. Generic ads paired with specific signals is a recipe for mediocre performance.
Consistent messaging between audience, ad, and landing page boosts conversion rates enough to offset PMax's lack of granular placement control. When everything aligns, the funnel works.
Test, Observe, and Iterate Without Constant Tinkering
Run meaningful tests: change one major audience factor at a time and allow at least 2–3 weeks before judging results.
PMax blends so many signals that over-editing audiences every few days keeps the algorithm in permanent "learning" mode. You never get stable performance because you never let optimization complete.
Set specific test timeframes before you start (e.g., "We'll evaluate after 14 days and 50+ conversions"). Document every change with dates so you can correlate performance shifts.
The Hard Restrictions You Still Control (Your Guardrails)
Even in an automated world, you need guardrails to keep the algorithm from going off the rails. Signals are guidance—but these are your real controls.
Negative Keywords: Keep your ads off "junk" searches. This is still one of the most effective ways to protect budget in PMax.
Brand/URL Exclusions: Make sure Google isn't sending traffic to your Terms & Conditions, Contact Us, or other low-value pages. Set these from the start.
New Customer Acquisition Settings: This toggle tells Google whether to value all customers equally or bid more aggressively for people who have never bought from you before. If you're running NCA mode, see the ROAS inflation note above—and make sure you know what you're looking at in your reports.
These guardrails work best when paired with smart pacing and budget discipline. Without them, even excellent signals can be overwhelmed by low-quality query volume.
Common Mistakes With PMax Audience Signals
Relying on Broad Interests Instead of Intent
The mistake: Building signals only around broad interests like "Business & productivity" or "DIY & home improvement." These reflect hobbies, not purchase decisions. You get lots of impressions and clicks, but your sales team gets garbage leads.
The fix: Rebuild key asset groups around high-intent custom segments and first-party converters. Review search term insights to identify phrases used right before conversions—those specific queries should seed new custom segments.
Before and after: A home services company used broad affinity audiences like "Home & garden enthusiasts" as primary signals. Lots of traffic, very few booked appointments. After rebuilding signals around specific custom segments ("water heater installation cost," "emergency plumber [city]") plus a customer match list of past clients, lead quality improved within two weeks and cost per booked job dropped by over 30%.
The fix:
Export past leads or customers from your CRM
Clean the data—remove spam, disqualified leads, refund customers
Segment by value (high-value buyers vs. general leads)
Re-upload as separate segments with clear labels
Prioritize cleaned lists in PMax audience signals
Simply fixing list hygiene—without changing anything else—can materially improve PMax campaign performance. I see this regularly in audits.
Ignoring or Misusing First-Party Data
The mistake: Leaving customer match lists and remarketing lists unused, or using bloated lists that mix good and bad leads with no value tiers. This confuses the algorithm about what a "good" customer looks like.
The PMax Audience Signals Audit Checklist
When PMax isn't performing, this is the five-step diagnostic I run in every account.
Step 1 — Identify Signal Drift
Signal drift happens when the machine starts optimizing for audiences that don't match your economics. Diagnose it by comparing who you intended to reach vs. who the machine is actually reaching vs. what your business actually needs. Drift is not a mistake—it's a natural outcome of a volume-maximizing system left without proper guardrails.
Step 2 — Evaluate First-Party Data Quality
Bad data poisons the system. If your customer lists are outdated, incomplete, or inflated with low-value users, the machine will optimize toward the wrong people. First-party data is the strongest signal you have and the easiest to misuse if you just upload any old list.
Step 3 — Map Signals to Actual Business Economics
Signals must reflect your margin realities, the household income of your target customers, your sales cycle length, your operational capacity, and your customer lifetime value. If your signals don't reflect your economics, the machine will optimize for outcomes you can't sustain.
Step 4 — Diagnose Misalignment Between Signals and Search Themes
Signals and Search Themes work together. Themes tell the machine what people are searching for. Signals tell the machine who those people might be. If they're misaligned, the machine gets confused—and confusion leads to wasted budget.
Step 5 — Rebuild Signals to Anchor the System to Profit
Most advertisers rebuild signals based on demographics or interests. I rebuild signals based on economics. You anchor the system to the customers who actually sustain the business—not just the customers who are cheapest to acquire.
Frequently Asked Questions
Do audience signals limit who sees my ads? No. Signals don't restrict your ads. They simply tell Google where to start, and then the algorithm expands to find more people who look like your best customers.
Are audience signals the same as targeting? No. Targeting is restrictive—it limits who can see your ads. Signals are guidance—they suggest who to start with, but Google expands beyond them.
What are examples of good audience signals? Clean first-party data (like past purchasers), competitor-based custom segments, and detailed demographic or interest layers. These help Google learn faster and waste less budget.
Can I control which signal drives conversions? Not directly. Google doesn't report which signal "worked." Your job is to provide the best inputs and let the machine optimize. Use the Insights tab to surface patterns over time.
How do audience signals connect to value-based bidding? Signals tell Google where to start. Value-based bidding teaches Google which conversions are worth more, so it can prioritize high-value customers. They're complementary tools—signals define your ideal audience, value-based bidding tells Google how much to pay to reach them.
Do small businesses really need audience signals? Yes—especially small businesses. Signals are particularly powerful for small budgets because they shorten the learning phase and reduce wasted spend during the early weeks when you can least afford to burn money on bad traffic.
What happens if I don't add any audience signals? PMax will still run, but it starts cold with no direction. The machine will explore whatever traffic it can find cheapest, which often means low-intent queries and broad, unqualified audiences. You'll spend more money, waste more time in the learning phase, and generate lower-quality leads before the algorithm figures out what a real converter looks like.
When You Should Get Expert Help With PMax Audience Signals
While many businesses can set up basic signals themselves, there are clear signs that indicate you should seek expert support rather than continuing to experiment alone.
Red flags that warrant outside help:
High spend with consistently low-quality leads
Strong Search or Shopping campaign performance but weak PMax results
Leadership demanding PMax usage without a clear strategy
Conversion tracking you don't fully trust
Months of "learning" with no improvement
A focused Google Ads audit can quickly reveal whether audience signals, tracking, or broader account structure are the main blockers. Often it's a combination—signals can't fix broken tracking, and good tracking can't fix absent signals.
From my perspective, getting Performance Max audience signals right is one of the fastest ways to cut wasted spend in new or underperforming PMax campaigns.
When I audit accounts, missing or poorly configured signals are often the first thing I fix. If you want to know if you have a true specialist managing your money, look here first; launching campaigns without these guardrails is precisely how inexperienced managers accidentally make $27,000 PMax mistakes by feeding a blank map to the algorithm.
How I work with clients:
Second-opinion audits: I review your account, identify where money is leaking, and give you a clear picture of what's working and what's not.
1:1 coaching calls: We get on a call and fix issues live in your account—no waiting for a report, no vague recommendations.
90-day build-and-train engagements: I build or restructure your campaigns and train you or your team to manage them profitably.
Audience signals are powerful, but they're just one piece of a larger system. Getting them right requires accurate tracking, clean data, aligned creative, and realistic expectations. When those pieces work together, PMax stops feeling like a gamble and starts delivering accountable results.
Ready to find out if your PMax audience signals are actually working? A structured audit can reveal exactly where your budget is going and whether your signals are guiding Google toward buyers or wasting spend on non-converters. Learn more about getting a second opinion on your Google Ads account.
