- Why Most AI Max Tests Waste Budget
- What Is AI Max Testing?
- How Do Google Ads Experiments Work for AI Max?
- How to Test AI Max Without Wasting Your Budget: 8 Steps
- AI Max Conversions vs Real Leads: Why More Conversions Do Not Always Mean More Business
- How Long Should an AI Max Test Run, and How Much Budget Do You Need?
- AI Max Testing Mistakes That Cost Australian Businesses Money
- How to Measure AI Max Campaign Performance
- AI Max Testing Checklist for Australian Businesses
- How DigitalHub360 Helps With AI Max Testing
- Final Thoughts
- FAQs
Most businesses judge AI Max by the number Google puts on the screen. Conversions rise and the report looks healthy. Nobody checks whether the extra enquiries were worth answering.
That gap is where test budgets disappear.
Google's own figures are encouraging. When AI Max launched in May 2025, Google reported that advertisers typically saw 14% more conversions or conversion value at a similar CPA or ROAS. Closer to home, Australian utility connection service MyConnect recorded 16% more leads at a 13% lower cost per action.
Those are averages and case studies, though. They say nothing about your account or your margins.
A quick disclosure: we manage Google Ads for Australian businesses and agencies, so we have a stake in this topic. This guide covers how to run a fair test and what to measure once it ends.
Why Most AI Max Tests Waste Budget
Most failed tests are set up badly rather than beaten by the algorithm:
- Conversion tracking counts page views or button clicks as wins.
- AI Max is switched on across the whole account at once.
- Exclusions are added after the test starts.
- The test ends early because the first week looked bad.
- Results are judged on raw conversions, not qualified leads.
Every one of these is a set-up problem you can fix before spending a dollar.
What Is AI Max Testing?
It is a controlled comparison between a Search campaign with AI Max switched on and the same campaign without it. Traffic is split between both versions, so any difference in results comes from AI Max itself.
Every test needs one primary metric. For most Perth service businesses, that is cost per qualified lead. For online stores, it is revenue.
Secondary signals, such as new search terms and the landing pages Google chose, explain why the result happened. Watch them, but never let them overrule the primary metric.
How Do Google Ads Experiments Work for AI Max?
Google's experiments tool lets you compare a change against your current setup before committing to it. For AI Max, Google offers a one-click experiment that splits traffic inside an existing campaign between AI Max on and AI Max off.
Two updates made testing fairer. On 20 August 2026, Google announced that AI Max experiments can now run with brand and location controls switched on. Multi-campaign A/B tests for budget and ROI target changes are also rolling out from September 2026, so some accounts may not have them yet.
How to Test AI Max Without Wasting Your Budget: 8 Steps
- Fix conversion tracking: Set qualified leads and completed sales as primary goals.
- Pick the right campaign: Choose one with steady conversion volume, not your weakest.
- Record a baseline: Save 30 days of cost per lead and a copy of the search terms report.
- Set guardrails first: Add negatives and brand exclusions before launch, and write text guidelines for generated copy.
- Split traffic 50/50: Change nothing else while AI Max testing is underway.
- Choose what you are testing: Google says the full feature suite averages 7% more conversions than search term matching alone, so it is a sensible default.
- Run it for four to six weeks: Ignore the first week while the system learns.
- Set a stop-loss rule: We recommend pausing if cost per qualified lead runs more than 30% above baseline after two weeks.
A test is only as honest as the baseline you measure it against.
AI Max Conversions vs Real Leads: Why More Conversions Do Not Always Mean More Business
As an illustration, over five weeks, the AI Max version of a campaign records 40 conversions against 32 for the control. On the dashboard, AI Max wins.
Then someone checks the enquiries. Only 14 of the AI Max leads were worth quoting, against 18 from the control.
Keywordless matching can attract researchers alongside buyers. A Joondalup physio might see clicks from students chasing course information, which look like engagement and never become bookings.
Frankly, this is where most AI Max testing goes wrong. Teams see a higher conversion count and scale a campaign that is quietly getting worse.
How Long Should an AI Max Test Run, and How Much Budget Do You Need?
Four to six weeks suits most accounts. Shorter tests rarely separate a real result from normal weekly swings.
In our experience, campaigns spending under about $2,000 a month seldom produce a clear answer. Sensible Google Ads budget optimisation means testing only where the data can support a decision.
Avoid testing across EOFY in June or the Black Friday to Boxing Day period, when seasonal spikes distort both versions.
AI Max Testing Mistakes That Cost Australian Businesses Money
- Testing many campaigns at once: You cannot tell which change caused the result.
- Ignoring generated ad copy: The Australian Consumer Law still applies to AI-written ads, and Ahpra rules apply to health practitioners.
- Skipping the landing page check: Final URL expansion can send paid clicks to old or thin pages.
- Declaring a winner too early: One good week is not a trend.
How to Measure AI Max Campaign Performance
Measuring how AI Max performs means looking past the conversion column:
| Metric | What it tells you |
|---|---|
| Cost per qualified lead | Whether AI Max finds better customers or simply more of them |
| Conversion rate | How well the new traffic turns into enquiries |
| Spend on AI Max queries | How much budget went to searches you never targeted |
| Search term relevance | Whether new queries match what you actually sell |
| Landing page mix | Which pages Google chose for each click |
Track phone calls too. Many Perth service businesses receive more calls than form submissions.
AI Max Testing Checklist for Australian Businesses
- Primary conversions set to qualified leads or sales
- One campaign with steady conversion volume
- A 30-day baseline saved
- Negatives and brand exclusions in place
- Text guidelines written
- A 50/50 split and a fixed test window
- A stop-loss rule agreed in writing
- Call tracking switched on
If you cannot tick at least six, fix those gaps before the test goes live.
How DigitalHub360 Helps With AI Max Testing
We plan and run AI Max tests for Australian businesses and for agencies under their own brand. Our Google Ads budget optimisation work starts with tracking and guardrails, and agency partners receive clear reports on AI Max campaign performance under their own logo.
When a test shows the page is the weak point, we build landing pages worth sending traffic to. If your campaigns only just moved across, start with our AI Max upgrade checklist.
Final Thoughts
Open your search terms report this week and filter for queries found by AI Max. Read the first fifty and count how many you would happily pay for.
Most accounts do not need more automation switched on. They need a cleaner test first. Set one up through Google Ads experiments and let qualified leads make the call.
Careful testing is slower than flipping a switch, and far cheaper than scaling the wrong result.
Agencies can send us one client account for a free AI Max test plan delivered under their brand.
Get In Touch→Frequently Asked Questions
What is AI Max testing?+
It is a split test comparing a Search campaign with AI Max on against the same campaign with it off. It shows whether AI Max lowers cost per qualified lead in your account.
How long should an AI Max experiment run?+
Four to six weeks suits most accounts. Ignore the first week while the system learns.
Does running an AI Max experiment cost extra budget?+
Not necessarily. The one-click experiment divides traffic within your existing campaign. The real cost is wasted spend when guardrails are missing.
What metric should decide whether AI Max stays on?+
Cost per qualified lead for service businesses, or revenue for online stores. Raw conversions can rise while lead quality falls.
References
- Google, Introducing AI Max for Search campaigns (6 May 2025):
https://blog.google/products/ads-commerce/google-ai-max-for-search-campaigns/ - Google, Make AI Max work for your business with new testing and planning tools (20 August 2026):
https://blog.google/products/ads-commerce/ai-max-testing-planning-tools/ - Google, We're upgrading Dynamic Search Ads to AI Max (updated 11 June 2026):
https://blog.google/products/ads-commerce/dsa-upgrade-to-ai-max-2026/ - Google Ads Help, One-click experiments for AI Max:
https://support.google.com/google-ads/answer/16450159 - Search Engine Land, Google adds new AI Max testing and planning tools:
https://searchengineland.com/google-adds-new-ai-max-testing-and-planning-tools-485566 - Australian Competition and Consumer Commission (Australian Consumer Law):
https://www.accc.gov.au/ - Australian Health Practitioner Regulation Agency (Ahpra):
https://www.ahpra.gov.au/