TL;DR
- What PMax is: One campaign covering every Google inventory type: Search, Shopping, Display, YouTube, Gmail, and Maps. Google's algorithm controls placement and audience selection.
- What you control: Asset groups, audience signals, search themes, budget, ROAS targets, and account-level negative keywords
- Key risk: PMax cannibalizes Search campaigns if you do not actively monitor Search Impression Share and set priority controls
- AI Agent role: Synter configures asset groups, signals, and search themes, then monitors cannibalization and search term data continuously
What Performance Max Actually Is
Google Inventory Types in One Campaign
Headlines Required Per Asset Group
of Google Ads Spend Goes to PMax (Industry Estimate)
Performance Max is Google's all-inventory campaign type. A single PMax campaign delivers ads across Search, Shopping, Display, YouTube, Gmail, and Maps from one setup. Google's algorithm decides which inventory to use for each impression, which audiences to reach, and how to allocate budget across placements in real time.
That is the source of both PMax's appeal and its controversy. In theory, Google's algorithm finds the best opportunity across all its surfaces to hit your conversion goal. In practice, advertisers lose granular control and visibility that they had with individual campaign types. The reporting is thinner than Search. The controls are fewer. The algorithm needs time and data to learn, which means the first weeks of a PMax campaign often look worse than what came before.
Despite the frustration, PMax is important. By 2026, Google has made clear that PMax is the direction of their ad product. Smart Shopping campaigns were migrated to PMax. Local campaigns merged into PMax. Advertisers who avoid PMax entirely are leaving inventory and audience reach on the table. The better approach is understanding exactly what you can and cannot control, then operating accordingly.
This guide covers the controls you actually have, how to configure asset groups effectively, how to guide the algorithm with audience signals and search themes, and how to detect and prevent the most common PMax failure mode: cannibalizing your own Search campaigns.
The Controls You Actually Have
Performance Max gives advertisers fewer levers than traditional campaign types, but the levers that exist matter. Knowing which controls are available prevents wasted effort trying to configure things that are not exposed:
Asset Groups (up to 100)
Asset groups are the core organizational unit inside PMax. Each group contains your creative assets (headlines, descriptions, images, videos) and the audience signals you want Google to use as starting guidance. You can create up to 100 asset groups per campaign.
Audience Signals
Audience signals tell Google which users to prioritize when learning. They are not hard targeting. Google can expand beyond them. But they accelerate the learning phase and reduce early-phase waste. Customer Match lists, custom intent audiences, and in-market segments are the strongest options.
Search Themes
Introduced in 2023, search themes let you provide keyword intent guidance to PMax. They are not keywords in the traditional sense. PMax does not match to them directly. Instead, they signal to Google what kinds of search queries are relevant to your goals.
Budget and ROAS Targets
You set a daily budget and, optionally, a target ROAS or target CPA. Google's bidding algorithm attempts to hit these targets while spending the full daily budget. Setting a ROAS target that is too aggressive can cause PMax to under-spend or restrict reach.
Account-Level Negative Keywords
PMax does not support campaign-level negative keywords. This is a significant limitation. Negatives must be applied at the account level, which means they affect all campaigns. Build your negative list carefully. A request to Google for PMax campaign-level negatives can sometimes be granted through your account rep.
What Is Off-Limits
Placement exclusions at the campaign level, keyword-level bidding, ad scheduling controls, and device bid adjustments are either absent or severely limited in PMax. You cannot control which specific YouTube videos, Display placements, or Gmail categories your ads appear on without contacting Google support. These are inside Google's black box.
Asset Group Strategy
The most impactful structural decision in Performance Max is how you organize asset groups. Multiple focused asset groups consistently outperform a single large asset group with all assets thrown in. Here is why and how to structure them:
When you create separate asset groups by product category, audience persona, or creative theme, Google can match the right creative to the right context. A single asset group forces Google to serve all creative across all audiences. Generic creative goes to high-intent users and vice versa. Segmented asset groups give the algorithm better material to work with.
| Segmentation Approach | When to Use | Example |
|---|---|---|
| Product Category | E-commerce with distinct product lines | One group for running shoes, one for trail shoes, one for accessories |
| Audience Persona | B2C with distinct customer segments | One group with CRM-matched buyers as signal, one with in-market prospects |
| Creative Theme | Multiple value propositions or seasonal offers | One group for price-focused creative, one for quality-focused creative |
| Geography | Different markets with distinct messaging needs | One group per major market with localized headlines and images |
Asset requirements per group are substantial. Google requires 15 headlines, 5 descriptions, at least one landscape image (1.91:1), one square image (1:1), one portrait image (4:5), and ideally a 15-second and a 30-second video. If you do not provide video, Google will auto-generate one from your images. Auto-generated videos typically underperform.
Google rates each asset as Low, Good, or Best. These ratings reflect how often that asset appears in winning combinations, not a quality judgment. An asset rated Low may simply be redundant with a Best-rated asset. Rotate out Low-rated assets after a few weeks and test new variations in their place.
Audience Signals: How to Guide the Algorithm
Audience signals are your primary mechanism for guiding PMax's learning phase. The better your signals, the faster Google finds your real customers, and the less budget gets consumed by exploration. Here is how to build a strong signal strategy:
Customer Match Lists from CRM
Upload your existing customer list as a Customer Match audience and attach it as a signal to your asset group. This is the highest-quality signal available. Google analyzes the characteristics of your real customers and finds users with similar attributes across its inventory. The larger and more recent the list, the better the signal quality.
Custom Intent Audiences
Build a custom intent audience using purchase-intent keywords relevant to your product. These are terms people actively search for when they are in a buying mindset. Custom intent audiences built from these terms give PMax a high-quality behavioral signal without requiring a large first-party data list.
In-Market Segments
Google's in-market segments group users who have shown recent purchase intent in a category based on their browsing and search behavior. Add relevant in-market segments as supplementary signals alongside your Customer Match and custom intent audiences.
Signals accelerate learning but do not restrict reach. Google may expand beyond your signals if it identifies higher conversion probability elsewhere. This is a deliberate design choice. PMax is built to find incremental conversions your signals would not have targeted. Monitor your audience insights panel to see where Google is actually converting and refine your signals accordingly.
Synter configures and monitors audience signals as part of PMax campaign setup. The agent syncs CRM lists to Customer Match, builds custom intent audiences from keyword performance data, and monitors the audience insights panel to detect when Google is converting in unexpected segments.
Search Themes
Search themes were added to Performance Max in late 2023. They give advertisers a way to provide keyword intent guidance directly to PMax without using traditional keywords. Understanding what they are and how to use them is now a core part of PMax strategy.
A search theme is a phrase that describes the kinds of searches relevant to your campaign. It is not matched directly to search queries the way a keyword is. Instead, it informs the algorithm about the intent signals you care about. Google uses search themes alongside your asset groups and audience signals to determine where your ads should compete.
You can add up to 25 search themes per asset group. The most effective approach is to derive them from your top-performing Search campaign keywords. If a keyword is generating high-quality conversions in your Search campaigns, adding a corresponding search theme to your PMax asset group signals that intent to the algorithm.
Search Theme Best Practices
- Use phrases, not individual words. "Buy running shoes online" gives more intent context than "running shoes."
- Focus on commercial and purchase-intent phrasing, not informational queries
- Revisit search themes monthly and update based on what your Search campaigns are converting on
- Match search themes to the specific asset group they belong to. Product-focused themes go in product asset groups, not mixed across everything.
Synter derives search themes from your Search campaign's top-converting keywords automatically. The agent reviews keyword performance data, identifies the highest-value intent signals, and populates search themes in the corresponding PMax asset groups through the Google Ads API.
The PMax Reporting Problem
PMax is one of the least transparent campaign types Google offers. Understanding what you can and cannot see is critical for operating it effectively rather than flying blind.
| What You Can See | What You Cannot See |
|---|---|
| Asset performance ratings (Low, Good, Best) | Full search term reports showing which queries triggered ads |
| Conversion volume and value by campaign | Individual placement-level performance (which YouTube videos, Display sites) |
| Audience insights panel (which segments converted) | Auction insight data at the campaign level |
| Partial search term insights (aggregated, not complete) | Device-level or time-of-day performance breakdowns |
| Asset group performance comparisons | Geographic breakdowns by city or region within PMax |
The reporting gap is real and it matters. Without search term data, you cannot confirm which queries are driving conversions. Without placement data, you cannot identify poor-quality Display or YouTube placements. Google argues that sharing this data would allow advertisers to over-constrain the algorithm and reduce its ability to find incremental conversions.
Synter compensates for this by pulling GA4 landing page data to understand which pages are receiving PMax traffic and converting, tracking Search Impression Share on high-value terms to detect cannibalization, and using holdout analysis to measure true incremental contribution. Holdout analysis pauses PMax for subsets of the account to isolate its effect.
PMax vs Search: Managing Cannibalization
The most common PMax problem is cannibalization. Performance Max operates on a higher campaign priority than standard Search campaigns. When a user's search matches intent covered by both a PMax campaign and a Search campaign, PMax wins the auction.
This matters most on branded terms and high-converting non-branded keywords where your Search campaigns have strong Quality Scores and efficient CPCs. PMax may step in, win those impressions at higher CPCs, and your Search Impression Share on those terms drops. Your branded CPC climbs. Your overall efficiency falls even though conversion volume looks stable on the surface.
How to Detect Cannibalization
- Monitor Search Impression Share on your branded terms and top-performing non-branded terms weekly
- Compare Search campaign CPC trends before and after PMax goes live. Rising CPCs on previously stable terms are a signal.
- Check Search Impression Share Lost (Budget) and Lost (Rank) for your Search campaigns after PMax is live
- Review the PMax audience insights panel for overlap with your existing Search audiences
The fix for cannibalization is account-level negative keywords that block PMax from bidding on branded terms and high-priority Search terms. Add your brand name, brand plus product variations, and any navigational queries to an account-level negative list. This forces those queries back to your Search campaigns where you have more control.
Synter monitors Search Impression Share against a configured baseline. When cannibalization is detected on priority terms, the agent flags the issue, identifies the affected terms from partial search term data, and adds them to the account-level negative keyword list through the Google Ads API.
Negative Keywords
Negative keywords in Performance Max work differently from every other Google Ads campaign type. There are no campaign-level negative keywords for PMax. All negatives must be applied at the account level, which means they affect your Search, Shopping, and Display campaigns as well.
Building a solid negative keyword list before shipping a PMax campaign is important. Once PMax is live, it will start bidding on anything Google deems relevant to your asset groups and signals. Without negatives, it will enter auctions for irrelevant queries, competitor brand terms, and informational searches that will never convert.
Starting Negative List Categories
Begin with competitor brand terms you do not want to bid on, irrelevant product categories adjacent to yours, informational query prefixes (how to, what is, definition of), geographic terms outside your target market, and any branded terms belonging to your Search campaigns.
Mining Search Term Data
PMax provides partial search term insights in the Insights panel. This data is aggregated and incomplete, but it surfaces queries that drove significant volume. Review it weekly and add irrelevant terms to your account-level negative list. Over time, this improves PMax query quality and reduces wasted spend.
Requesting Campaign-Level Negatives
For accounts with significant spend, Google sometimes grants access to a PMax campaign-level negative keyword tool through account representative requests. This is not available by default but worth requesting if your account has a dedicated rep. It allows more surgical exclusions without affecting other campaigns.
Synter and PMax Negative Keywords
Synter's Google Ads agent mines the partial search term insights report weekly, identifies queries that do not match your target intent, and adds them to the account-level negative keyword list through the Google Ads API. The agent tracks negative keyword additions over time and surfaces patterns in irrelevant query categories that inform broader campaign structure decisions.
Frequently Asked Questions
How is Performance Max different from other Google Ads campaigns?
Performance Max is Google's all-inventory campaign type. A single PMax campaign delivers ads across Search, Shopping, Display, YouTube, Gmail, and Maps simultaneously. Google's algorithm decides which inventory to use and which audiences to reach based on your asset groups, audience signals, and conversion goals. Traditional campaigns run on a single inventory type with advertiser-controlled targeting.
What audience signals should I add to Performance Max?
Customer Match lists from your CRM are the highest-quality signal you can provide. They tell Google exactly who your best customers are. Custom intent audiences built from purchase-intent keywords come second. In-market segments add another layer. Signals accelerate the learning phase, but Google may expand beyond them. Providing high-quality signals reduces the time and cost of the learning period.
Why can't I see which search terms triggered my PMax ads?
Google limits PMax search term reporting to protect their algorithm. You can access partial search term insights in the Insights panel and asset performance ratings (Low, Good, Best) at the asset level, but full keyword-level data is not available. Synter compensates by monitoring GA4 landing page data, tracking Search Impression Share on key terms, and running holdout analyses to measure PMax contribution.
How do I prevent Performance Max from cannibalizing my Search campaigns?
Use campaign priority settings and monitor Search Impression Share on your branded and highest-value terms. When PMax steals impressions from Search, your branded Search Impression Share drops. Add account-level negative keywords to block PMax from bidding on terms your Search campaigns own. Synter detects cannibalization by tracking Impression Share trends and alerts when thresholds are crossed.
Can AI Agents manage Performance Max campaigns?
Yes. Synter's Google Ads agent configures asset groups, uploads audience signals, sets search themes, and establishes budget and ROAS targets through the Google Ads API. The agent monitors PMax performance against Search campaigns, detects cannibalization when Search Impression Share drops on priority terms, and mines partial search term data to expand account-level negative keyword lists.