When AI Max Search Falls Short: Alternative Strategies For Advertisers

Google is aggressively pushing AI Max for Search as the next evolution of PPC, a powerful AI layer promising “14% more conversions or conversion value at similar cost per acquisition or return on ad spend.” For busy PPC marketers and growth-focused CMOs, it sounds like a magic bullet. But, there’s just one problem: the independent data flatly contradicts this.

New, extensive analyses from third-party advertising experts, released in November 2025, paint a starkly different picture. The consensus? AI Max isn’t just failing to meet expectations; it’s actively underperforming traditional, advertiser-controlled match types, in some cases dramatically. 

One 4-month test found AI Max delivered conversions at a 90% higher CPA than Phrase Match. Another analysis of over 250 retail campaigns found AI Max delivers 35% lower ROAS. This creates a critical dilemma for advertisers: you’re being pushed to adopt a “black box” that new evidence suggests may be burning your budget.

What Independent Tests Reveal About AI Max Performance

Two recent analyses have sent a chill through the PPC community. In early November 2025, advertising strategy consultancy Smarter Ecommerce published an analysis of over 250 retail campaigns. Their findings were conclusive:

• 35% Lower ROAS: AI Max consistently underperformed other match types within the same campaigns, delivering conversions at a 35% lower ROAS.
• Higher CPA & Lower AOV: The problem was twofold. Cost per conversion was significantly higher, while the Average Order Value (AOV) of the conversions it did find was “notably lower.”

This combination of paying more for less valuable customers is the exact opposite of the efficiency AI promises.

Paid Ads Specialist’s 4-Month Test: “I’m Done with AI Max”

In a separate analysis, also published on November 6, 2025, Paid Ads expert Xavier Mantica shared what he called brutal data from a four-month head-to-head test. The results are a clear visualization of the advertising match type ROAS comparison:

Match Type	Cost Per Acquisition (CPA)
AI Max	$100.37
Phrase Match Close Variants	$97.67
Exact Match Close Variants	$61.65
Exact Match	$52.69
Phrase Match	$43.97

AI Max lost to every other match type, including the close variants that advertisers often criticize for poor performance. It cost 90% more to get a conversion with AI Max than with a well-structured Phrase Match keyword.

Google’s Claims vs. Market Reality

This data is a directly challenges Google’s official claims of a 14% lift. It’s important to note that Google’s own fine print states it “offers no performance expectations for retail campaigns” using AI Max. Furthermore, an independent LinkedIn poll created by Google Ads specialist Adriaan Dekker showed that only 16% of advertisers reported good performance from the feature.

This signals a massive disconnect between Google’s internal, aggregated data and the granular, on-the-ground reality that performance-focused marketers are experiencing.

5 Reasons AI Max Fails Performance Marketers

When a new feature underperforms this badly, advertisers need to know why. The data suggests several structural flaws in how AI Max operates.

1. It’s Not “New AI,” It’s “Rebranded Broad Match” 

Mantica’s analysis concluded that AI Max is essentially Google’s attempt to rebrand Broad Match, a setting many seasoned advertisers distrust for its tendency to burn budgets on irrelevant searches. It’s a black box that strips away keyword-level control and turns your carefully selected keywords into mere suggestions for the AI.

2. The Myth of Efficient Incremental Conversions 

The common defense for this type of expansion is that incremental or on-top conversions will naturally cost more. While true, a 35-90% performance drop isn’t an incremental cost; it’s a sign of a failed strategy. The data shows the AI is finding low-quality, low-AOV traffic that advertisers had likely already optimized against.

3. It Actively Targets What You Excluded: Competitors and Junk 

Why machine match fails PPC often comes down to this: the AI doesn’t understand your business strategy. Mike Ryan, Head of Ecommerce Insights at Smarter Ecommerce, shared that AI Max aggressively targeted competitor brand terms, accounting for 69% of total impressions, a tactic many brands purposefully avoid due to high cost and low intent.

4. The Search Partner Network Problem 

Analysis shows AI Max generates disproportionate impression volumes across Google’s Search Partner Network (SPN). This is a critical issue because the SPN is notoriously less transparent and consistently underperforms Google Search proper, delivering 37% lower ROAS, according to Ryan. AI Max is automatically pushing your budget into these lower-quality, higher-risk placements.

5. Account Cannibalization and Chaos 

AI Max doesn’t exist in a vacuum. It creates significant, confusing overlap with PMax, Dynamic Search Ads, and even your existing Broad Match keywords. This turns the account into a tangled system of automations fighting over the same traffic, leaving advertisers with little clarity and even less control.

Why AI Max Is a Magnet for Click Fraud

As a click fraud protection company, we see a deeper, more dangerous problem: less control is an open invitation for fraud.

The “irrelevant traffic” and “low-AOV conversions” reported aren’t just low-intent humans. In many cases, it’s non-human, invalid traffic.

How Keywordless Targeting Invites Bots

When you remove keyword control, you remove your primary layer of defense. Vague, semantic matching based on landing page content is trivial for bots to trigger. Sophisticated botnets and click farms don’t need to guess your exact keywords. They just need to mimic a relevant user profile or query, and the AI serves them your ad.

The 69% impression share on competitor terms? Much of that is likely competitor bots and ad scrapers, not just curious customers.

The Black Box Placement Danger

The biggest risk is the forced expansion onto the Search Partner Network. The SPN is a well-known haven for click fraud precisely because it lacks transparency. Fraudsters set up low-quality “parked domain” or “made for ads” sites, join the SPN, and use bots to click the ads.

When AI Max pushes disproportionate volume to these partners, it’s pushing your budget directly into the hands of fraudsters. This is the exact same vulnerability we see in Performance Max campaigns.

The ClickGuard Verdict: The 35% ROAS drop is a clear signal of massive budget drain from invalid sources. You’re not just paying more for bad customers; you’re paying for clicks that were never human to begin with.

When Does Machine Match Work? When Does It Fail?

To be clear, AI-driven automation isn’t always bad. But it’s a specific tool for a specific job. The problem is that AI Max is being marketed as a one-size-fits-all solution. Here is a simple breakdown of when to consider automated expansion vs. when to maintain control.

Machine Match (AI Max, PMax) Might Work If…

• Your goal is maximum reach: You are prioritizing brand awareness, top-of-funnel traffic, and query discovery over bottom-line efficiency.
• You have a very large budget: You can afford the waste of an 8-week (or longer) learning period and the cost of the AI finding low-quality traffic.
• You’re in a brand-new market: You have no historical data and need the AI to discover new query themes and search patterns.
• You sell simple, high-volume goods: Your product is B2C, has a short sales cycle, and benefits from broad, semantic matching.
• You have no brand safety concerns: You don’t mind your ads appearing for competitor terms, tangential topics, or low-intent queries.

Traditional Match Types (Phrase/Exact) Will Outperform If…

• You have strict ROAS or CPA targets: You must maintain profitability and cannot afford the 35-90% drop in efficiency seen in independent tests.
• You have a limited or fixed budget: Every dollar counts. You must focus your spend on high-intent searches, not AI exploration.
• You’ in a mature, optimized campaign: You have years of data and have already optimized against the low-quality traffic AI Max will discover.
• You’re in a niche or B2B industry: You must control your targeting to avoid irrelevant, consumer-grade traffic and high-value junk leads.
• You must protect brand integrity: You absolutely need to exclude competitor brand terms, negative-intent queries, or other topics that dilute your brand.

What to Do Instead of AI Max

For 90% of advertisers, the data shows that AI Max Search ads are not the right move. Here is a practical framework for moving forward.

1. Re-Master the Modern Match Types (Your Foundation) 

Instead of ceding control, double down on what works.

• Phrase Match: This is your new workhorse. It now uses AI to capture intent and relevance but keeps you in control of the core theme. The $43.97 CPA in Mantica’s test proves its power.
• Exact Match: This is your scalpel. Use it for your high-intent, bottom-of-funnel, and branded money terms.
• Controlled Broad Match: If you want expansion, use standard Broad Match paired with a Smart Bidding strategy. This lets Google’s AI optimize for signals while you still control the keyword seed and, most importantly, your negative keyword lists.

2. A Safe AI Max Testing Protocol (If You Must Test)

If you are curious, don’t apply AI Max to your core, high-performing campaigns. Instead:

• Isolate it: Use Google’s Experiments feature to run a true A/B test with a 10-20% budget split.
• Set a tight budget: Google recommends a $50/day minimum and an 8-week learning period. This is an expensive test.
• Monitor aggressively: Watch the new reports like AI Max expanded matches and AI Max expanded landing pages. Be prepared to add hundreds of negative keywords.

3. Implement Proactive Protection (The Non-Negotiable Step) 

Before you test any black-box system—whether it’s PMax or AI Max—you need independent validation. A third-party click fraud solution is the layer that tells you who is actually clicking. Without it, you’re guessing: Are those clicks human? Is that spike in growth legitimate traffic or automated behavior? Is the 35% lower ROAS a performance gap… or simply wasted budget from invalid clicks?

When you don’t monitor traffic quality, you can’t distinguish between algorithmic inefficiency and outright invalid activity. That makes your test results unreliable, your optimizations misguided, and your conclusions potentially wrong.

Conclusion: Control, Not Just Automation, Is the Future of PPC

The latest independent data is clear: AI Max is, in its current form, a step backward for performance-driven advertisers. It’s a black box that trades your strategic control for vague promises, all while exposing your budget to massive inefficiency and invalid traffic.

The future of PPC isn’t a fully-automated, machine-only system. The winning strategy is human + machine. Use AI where it’s strong, like in Smart Bidding, but maintain human control over your strategy, your targeting, and, most critically, the validation of your traffic.

Don’t let your ad budget become a costly training ground for Google’s AI.