Google Performance Max and AI Max: The 2026 Playbook for Singapore Advertisers

Google Ads in 2026 runs on two AI engines, and most advertisers misuse both. Performance Max (PMax) automates reach across every Google surface. AI Max layers machine learning onto standard Search campaigns. Used with the right structure and guardrails they expand results. Switched on blindly they quietly burn budget. This playbook is the deep version: the readiness checklist, the honest performance data, the setup methods, and the daily monitoring that separate a win from a write-off.

It builds on our Google Ads strategy and what is SEM guides.

PMax vs AI Max: The Real Differences

Most advertisers should run both alongside traditional Search. See Google Ads vs Facebook Ads for channel-level choices.

AI Max: The Three Features and Their Risk

AI Max is three independently toggled features, at campaign or ad group level. Know the risk of each before you enable it.

Control tip: Final URL expansion and text customisation carry the most risk. Use landing page exclusions and term exclusions so AI never sends traffic to the wrong page or writes off-brand or non-compliant copy.

Performance Reality: Set Expectations Honestly

Google’s case studies show dramatic wins, but independent data is far more mixed. Go in with realistic expectations, not the headline numbers.

Takeaway: Google features the top 1% of outcomes. Assume a realistic single-digit to low double-digit lift, and only if your account meets the readiness bar below.

Pre-Test Readiness Checklist

Do not enable AI Max or scale PMax until these are true. This is the difference between AI that learns and AI that wastes.

• Conversion tracking with deduplication and Enhanced Conversions live.
• Smart Bidding running (Target CPA or Target ROAS preferred over Maximize-only).
• At least 30 conversions per month, ideally 100 or more.
• Adequate budget. Google’s minimum is around $50 per day, but real-world testing points to roughly 15x your target CPA per day (often around $750) to absorb exploration spend.
• A history of broad match success in the account.
• Comprehensive negative keyword lists in place.

When to Use and When to Avoid AI Max

Three Ways to Launch (Use the Experiment)

Option 1: New campaign

Build a fresh Search campaign, set conversion goals and Smart Bidding, set an adequate budget, toggle AI Max on, then add brand and landing page exclusions.

Option 2: Enable on an existing campaign

Verify Smart Bidding is active, toggle AI Max at campaign level, choose features, and monitor closely through a 2 to 4 week learning period.

Option 3: Experiment (recommended)

1. In Experiments, select your control Search campaign.
2. Create a custom experiment with a 50/50 split.
3. Enable AI Max only in the treatment arm and match daily budgets.
4. Run a minimum of 4 weeks (about 2 weeks learning plus 2 weeks evaluation).
5. Compare conversions, CPA or ROAS, and conversion rate, then scale the winner.

Feature Rollout Strategy

Daily Monitoring Protocol (First Two Weeks)

1. Download the search terms report filtered to the AI Max match type.
2. Flag three red patterns: competitor brand terms eating budget, queries missing intent, and high impressions with zero conversions.
3. Add negative keywords immediately.
4. Review which headlines served on converting versus wasted spend.
5. Check the landing pages AI selected if URL expansion is on, and exclude poor ones.

Network tip: Exclude the Search Partner Network by default, since it commonly delivers 30%+ lower ROAS under AI Max. Add it back only if it proves out.

Exclude these landing pages: Help and support pages, FAQs, pages without conversion tracking, old test pages, and pages meant for other regions.

Performance Max Structure That Actually Works

PMax results are decided mostly by structure. The single biggest mistake is one giant asset group. Build for relevance.

• Create themed asset groups by product line, service, or audience, each with its own creative and signal.
• Use search themes to steer each group. As of 2026 you can add up to 50 themes per asset group, up from 25.
• Feed strong audience signals, especially Customer Match lists, and use the new audience exclusions.
• Provide diverse creative in all sizes, including video, and watch the Ad Strength indicator.
• Start bidding on Maximize Conversions or Value, and wait 2 to 3 weeks before applying strict CPA or ROAS targets.
• Review the channel breakdown, exclude weak Display placements under Where ads showed, and routinely add negatives from search query data.

For e-commerce, connect your feed and follow the Google Shopping optimisation guide.

Portfolio Allocation Framework

Pause or Discontinue Triggers

• CPA stays 30%+ higher than traditional Search after 4+ weeks with no improving trend.
• Nearly all impressions generate zero conversions.
• Search terms stay irrelevant despite aggressive negatives.
• Spend accelerates without matching conversions.
• Management overhead exceeds running a manual Search campaign.

Common Mistakes

• One giant asset group instead of themed groups.
• Enabling AI Max account-wide before testing.
• Undersized budget (the $50 minimum is rarely enough).
• Weak negative keyword lists with aggressive AI expansion.
• URL expansion or auto-assets enabled without review or guidelines.
• Including the Search Partner Network by default.
• Judging results before the 2 to 4 week learning period ends.

Cost Context for Singapore

Singapore is a high-CPC market. For planning ranges see Google Ads cost and Google Ads price in Singapore. Pair campaign work with a regular SEM audit.

Run PMax and AI Max the Right Way

MediaPlus is a Google Partner agency that structures and manages PMax and AI Max through our SEM and PPC services. Get in touch for a free account review.

Related Reads

• Types of Google Ads
• Google Ads strategy
• Google Ads optimization checklist
• Google Shopping optimization guide
• SEM audit checklist
• Conversion rate optimization