Google Ads AI Max Campaign Experiments – Strategy, Setup, and Success

What is “AI Max for Search,” and why run Google Ads AI Max Campaign Experiments?

AI Max for Search is Google’s new, campaign-level upgrade that layers two AI pillars on top of a regular Search campaign: Search term matching (to expand qualified query coverage using broad and keyword-less matching) and Asset optimization (to auto-tailor ad text and landing-page selection at scale). It’s built to optimize in real time while keeping brand-safety and reporting controls inside the familiar Search workflow.

Why experiments exist. Google has added a built-in experiment flow that diverts a percentage of traffic and budget within the same campaign instead of forcing you to duplicate campaigns. The goal is simple: faster reads, fewer setup mistakes, and shorter learning periods. When the test ends, you can auto-apply results if they’re favorable or apply them manually.

At a glance – the two AI pillars you’ll test

Pillar	What it does	Key controls & reports
Search term matching	Expands beyond your keywords with broad and keyword-less tech to capture relevant queries you’d otherwise miss.	Toggle at ad group level; brand controls at campaign and ad-group levels; Search terms report adds an AI Max match-type and “Search terms and landing pages from AI Max” view. (Google Help)
Asset optimization	Uses text customization and Final URL expansion to assemble the most relevant copy + destination per query.	Toggle at campaign level; manage URL inclusions/exclusions; review in Asset and Landing pages reports.

Why run Google Ads AI Max Campaign Experiments now

• Speed with guardrails. The experiment keeps control and treatment in a single campaign, so ramp time drops, sync errors fade, and reads arrive sooner. If the uplift is real, you can apply the winner at the click of a button.
• Modern reach, measurable impact. AI Max can add incremental qualified queries that your current keyword set does not catch. The experiment lets you measure that lift against the KPIs that matter, not just impressions.
• Transparency you can act on. Reporting adds AI-specific match type, query→headline→URL combinations, and clearer asset/landing-page views so you can see why a match occurred and tune controls without guesswork.
• Less ops overhead. You no longer need to clone campaigns or babysit two objects for weeks. The experiment slot handles the split and the post-test apply path.

Context from the field. Industry coverage notes that Google’s new AI Max experiment type removes the need for duplicate campaigns and fast-tracks testing through a built-in split inside the campaign shell. That makes “try before you roll out” practical for busy accounts.

Where AI Max fits in your Search stack
AI Max sits firmly in the Search lane. It is not Performance Max. Think of it as a Search boost that combines reach expansion and creative automation, backed by new controls like brand safety and locations of interest. That combination is exactly what the experiment is designed to validate in your data, with your economics.

Quick eligibility checklist before you test

• Your campaign should not use shared budgets or portfolio bidding.
• Don’t have legacy features like old text customization, brand inclusions/exclusions, or ad-group location inclusions already enabled in that same campaign.
• Display targeting in a Search campaign disqualifies this experiment flow.
  These limits come from Google’s official experiment documentation.

Pro tip for evidence-based rollouts. If you need stronger statistical power or run lower volume, pair AI Max experiments with Google’s Campaign Guidance (it shows an Experiment Power score and recommendations to reach significance faster).

How Google Ads AI Max Campaign Experiments work (the new built-in flow)

The short version: you test AI Max inside the same Search campaign by splitting traffic between a control (AI Max off) and a trial (AI Max on). This built-in flow cuts duplication work, reduces sync errors, and speeds learning. When the test ends, you can auto-apply the winner or apply changes manually. See Google’s help page and official blog for the latest details.

What the trial arm turns on by default (and where to control it)

Feature (AI pillar)	Default in trial	Control level	What it actually changes	Where you’ll see it
Search term matching	On	Ad group	Expands reach beyond your keywords using broad + keywordless tech	Search terms report adds AI Max match type + “source” column; combined view shows search term → headlines → URL
Asset optimization (Text customization + Final URL expansion)	On	Campaign (toggle each)	Builds intent-aligned copy and can route to a more relevant landing page	Asset and Landing pages reports; “Selected by” column shows AI-chosen URLs

Step-by-step: create a Google Ads AI Max Campaign Experiment

1. In Google Ads, open Experiments.
2. Click + → Campaign features and settings → choose AI Max for Search campaigns.
3. Pick your traffic split and duration.
4. Confirm defaults for the trial arm. Optionally disable Search term matching at the ad-group level or disable Asset optimization at the campaign level.
5. (Optional) Add brand controls, locations of interest, URL inclusions/exclusions for extra precision.
6. Decide whether to auto-apply if results are favorable at the end.
7. Launch the experiment and monitor in the Experiments tab.

Where results show up (and what to read first)

• Experiments tab → expanded Experiment summary for your head-to-head read.
• Search terms report → filter Match type = AI Max and use Search terms & landing pages from AI Max to see the exact term → headline(s) → URL combinations that served.
• Keywords report → two aggregate rows show Total: AI Max expanded matches and Total: AI Max landing page matches to quantify incremental lift beyond your keyword list.
• Asset report → review Automatically created entries and prune misaligned lines; Landing pages adds a Selected by column to verify routing.

Auto-apply vs manual apply

You have three ways to adopt the winner after the test:

• Auto-apply if results are favorable (set it during creation).
• Click Apply from the Experiments table after the test ends.
• Toggle AI Max in Campaign settings if you prefer direct control.

Setup limitations that can block this flow

• The campaign targets the Display Network.
• You use Portfolio bid strategies.
• You use Shared budgets.
• You already have legacy features enabled (older text customization, brand inclusions/exclusions, ad-group location inclusion) in that campaign shell.
  These limits apply to the built-in experiment flow.

Choosing a traffic split

Split option	When to use it	Trade-offs
50/50	You want the fastest read and can tolerate equal exposure to the trial.	Strong power and speed. Larger swing if the trial underperforms.
70/30	You want a gentler ramp while preserving performance stability.	Slower to significance. Useful for sensitive CPAs or seasonality.
90/10	You need a smoke test first.	Very slow to significance. Use only when risk tolerance is low.

Reports and trade coverage point to a 50/50 split as the default pattern for AI Max experiments. Adjust only if budget volatility is a concern.

Why this design speeds learning

Keeping control and trial in the same campaign shortens ramp since budgets, histories, and settings stay aligned. You avoid the typical drift that hits cloned campaigns. Google’s docs explicitly call out faster results, fewer setup errors, and a shorter learning period as benefits of this approach.

Google Ads AI Max Campaign Experiments vs Performance Max vs Broad Match – what changes, what stays, and what matters

AI Max for Search is an upgrade layer for Search campaigns, Performance Max is a separate, multi-channel campaign type, and broad match is a keyword match option you can use with Search. Those differences shape how your Google Ads AI Max Campaign Experiments should run and how you read the results.

The high-level comparison

Dimension	AI Max for Search	Performance Max	Broad match (in Search)
What it is	Feature suite added to a Search campaign to expand matching and optimize assets/URLs	Separate goal-based campaign that runs across Search, Shopping, YouTube, Display, Discover, Gmail, Maps	Keyword match type or a campaign-level setting that converts all keywords to broad
Where ads serve	Search only	All Google channels above	Search (within your Search campaigns)
How reach expands	Search term matching uses broad + keyword-less tech	AI distributes budget across channels and surfaces	Matches to related meaning of your keywords
Creative & landing pages	Asset optimization with text customization and Final URL expansion	Asset Groups combine text, images, video, feed; landing pages via URL expansion	Uses your RSA assets and landing pages as configured
Controls you get	Brand inclusions/exclusions, locations of interest, URL include/exclude	Channel insights, asset-group reporting, search-terms reporting, negative keywords, device/age targeting (rolling out)	Standard Search controls plus negatives; optional broad match campaign setting
Reporting you read	New AI Max match type in Search terms, Keywords, Assets, Landing pages	Newer channel-level views, search-terms for PMax, asset-group reporting	Standard Search terms and keyword reports
Great for	Proving incremental query coverage and conversion lift inside Search with guardrails	Scaling across channels when you have creative breadth and feed depth	Capturing more intent within Search when paired with Smart Bidding

Why AI Max is not Performance Max (and why that matters in experiments)

AI Max lives inside Search. It turns on two pillars – Search term matching and Asset optimization – then lets you test them with a built-in control vs trial split. You keep campaign context identical which reduces setup drift and speeds learning. That makes your Google Ads AI Max Campaign Experiments precise and attributable.

Performance Max operates across channels by design. It optimizes toward your goal and decides which surfaces deserve budget. In 2025 Google added channel performance, search-terms reporting for PMax, and asset-group insights to address transparency gaps. These help you compare channel contribution and fine-tune assets, yet they measure a different scope than an AI Max Search experiment.

Where broad match fits relative to AI Max

Broad match remains a keyword matching option in Search. You can also flip a campaign-level Broad match keywords setting that converts all keywords to broad when using conversion-based Smart Bidding. AI Max’s Search term matching uses broad and keyword-less methods to find related intent beyond your list which is broader than flipping match types alone.

Reporting differences you’ll feel during Google Ads AI Max Campaign Experiments

• AI Max, in Search: You get an AI Max match type inside Search terms plus Keyword, Asset, and Landing pages views with “Selected by” to show AI-chosen URLs. This visibility lets you verify query → headline(s) → URL combinations during and after the test.
  
• Performance Max: You now have channel-level views and search-terms access for PMax, plus deeper asset-group reporting. That is ideal for multi-surface scale tests but it is not an apples-to-apples read against a Search-only AI Max experiment.

When to test AI Max vs lean on PMax vs upgrade to broad match

• Choose AI Max experiments when you want Search-only rigor, brand guardrails, and a quick read on incremental query coverage without rewriting your structure. This is perfect when exact/phrase growth stalls or you need reach without losing control.
• Lean on Performance Max when you must scale across surfaces like Shopping + YouTube + Discover alongside Search with unified bidding and creative. Use the new reporting to validate where spend flows and which assets move the needle.
• Upgrade to broad match inside Search if you’re comfortable letting Smart Bidding interpret related meaning and you want wider intent capture without the full AI Max feature set. You can run Google Ads AI Max Campaign Experiments later to compare this baseline with AI Max’s broader matching plus creative/URL automation.

Practical guardrails as you compare

• Brand safety first. If you test AI Max, use brand inclusions/exclusions and URL include/exclude rules to keep experiments disciplined while you explore new queries and landing paths.
• Mind scope creep. A PMax test can win on total conversions because it taps more surfaces. That can be great for revenue yet it won’t answer whether Search alone improved via AI Max. Keep tests scoped to the question you need to answer.
• Read the right reports. In AI Max experiments, start with Experiments → Summary, then drill into Search terms (AI Max) and Landing pages. In PMax, review channel performance, search-terms, and asset-group views before judging winners.

Prerequisites and limitations before you start Google Ads AI Max Campaign Experiments

Before you switch on AI Max inside an experiment, confirm eligibility and controls. Google documents a few blockers that stop the built-in experiment flow. You should also understand which settings AI Max upgrades or requires so you avoid conflicts later.

Eligibility – what must be true before you create the experiment

• Campaign type: Your campaign must be Search-only. If the campaign targets the Display Network, you cannot use the built-in AI Max experiment.
• Budgets and bidding: Experiments do not run if the campaign uses a Shared budget or is part of a Portfolio bid strategy. Move the campaign to its own budget and bidding strategy first.
• Legacy feature conflicts: The built-in experiment cannot start if the campaign already has older text customization (formerly “Automatically created assets”), brand inclusions/exclusions, or ad-group location inclusion configured outside AI Max. These are now managed by AI Max and must not exist as legacy toggles.

Availability and rollout context

AI Max for Search is available globally and can be managed in Google Ads, Ads Editor, Search Ads 360, and the Google Ads API as part of the current rollout.

Required and upgraded settings you should know

AI Max is a feature suite. Turning it on changes where several controls live and how they behave. Plan your experiment with these in mind.

Setting	Where it lives under AI Max	Key rule you must respect
Search term matching	Campaign toggle on first, then per ad group you can keep it on/off	You must enable it at campaign level to manage it at ad-group level.
Text customization (formerly “Automatically created assets”)	Campaign → Asset optimization	Enabled by default when AI Max is on. It upgrades prior legacy behavior into AI Max.
Final URL expansion	Campaign → Asset optimization	Requires Text customization. If you turn off Text customization then Final URL expansion turns off.
Pinned RSA assets	Responsive Search Ads behavior	Pinning is not respected when Final URL expansion is on or when URL inclusions are used. Turn those off if you must enforce pins.
Brand settings (inclusions/exclusions)	Campaign level for exclusions and Campaign/Ad group for inclusions	These controls are part of AI Max. Avoid legacy versions in the same campaign.
Locations of interest	Ad group level	Helps guide keywordless matches toward geo intent. Plan it before launch.
URL controls	Campaign-level exclusions and Ad group-level inclusions	Use exclusions to block pages. Use inclusions to force landing on specific pages during matching.

Built-in experiment defaults and scope

When you create the Google Ads AI Max Campaign Experiment, Google turns on Search term matching and Asset optimization for the trial arm by default. You can disable Search term matching at the ad-group level and Asset optimization at the campaign level before launch. Your changes apply only to the trial arm.

Reporting prep – what your team needs access to

Make sure your team can read the new reporting surfaces before you launch. The Search terms report adds an AI Max match-type and a combined view of search term → headlines → URL. The Keywords report adds two AI Max totals. Asset and Landing pages reports expose AI-selected assets and URLs.

Practical preflight checklist

Use this as a quick gate before you hit “Create experiment.”

• The campaign is Search-only and not in a shared budget.
• The campaign is not in a portfolio bidding strategy.
• Legacy text customization, brand inclusions/exclusions, or ad-group location inclusion are removed from the old locations.
• You understand that Final URL expansion requires Text customization and may ignore pinned RSA assets.
• Brand lists, locations of interest, and URL include/exclude rules are defined in AI Max panels.
• Stakeholders know where to read the Experiments → Summary view and the AI Max reporting pages.

Set up a Google Ads AI Max Campaign Experiment – step by step

Goal: create a head-to-head test inside one Search campaign that splits traffic between a control (AI Max off) and a trial (AI Max on). This built-in flow reduces setup error and shortens learning time.

Launch the built-in Google Ads AI Max Campaign Experiments flow

1. In your Google Ads account, open Campaigns → Experiments.
2. Click + and choose Campaign features and settings.
3. Select AI Max for Search campaigns.
4. Confirm the experiment details (traffic split, duration).
5. Review the trial defaults. By default, AI Max enables Search term matching and Asset optimization for the treatment arm. You may disable Search term matching at the ad-group level and disable Asset optimization at the campaign level before launch.
6. (Optional) Add URL exclusions at the campaign level and URL inclusions at the ad-group level to shape landing-page behavior during the test.
7. Choose whether to auto-apply the experiment if results are favorable, or plan to apply manually after review.

Know the switches you will touch (and where they live)

Control	Where you’ll find it	What to remember
Search term matching	Enable at campaign level, then toggle per ad group	Must be on at campaign level to manage at ad-group level. Useful when you want some ad groups to stay keyword-only.
Text customization	Campaign → AI Max → Asset optimization	On by default when AI Max is on. Powers tailored copy from your domain, existing ads, and ad-group context.
Final URL expansion	Campaign → AI Max → Asset optimization	On by default and requires Text customization. If on, pinned RSA assets may be ignored when a more relevant URL is chosen.
URL exclusions	Campaign → AI Max	Blocks specific pages from being selected during Final URL expansion. AI Max must be enabled.
URL inclusions	Ad group → AI Max	Forces traffic to only the URLs you specify for that ad group, with special behavior when combined with Final URL expansion.

Reading and applying results (what to line up before launch)

• Where to read: Start in Experiments → Experiment summary for the head-to-head view. Then drill into Search terms, Keywords, Assets, and Landing pages for AI Max-specific insights (including AI Max match type, search term → headlines → URL combinations, and Selected by for landing pages).
• How to apply: You can pre-select Apply if results are favorable during setup, or click Apply from the Experiments table after the test ends. You can also enable AI Max directly in Campaign settings post-test.

Alternate paths (only if you’re building new or retrofitting)

• Create a new Search campaign with AI Max on: During campaign creation, toggle AI Max on the dedicated AI Max page and pick features; ad-group controls come after AI Max is enabled.
• Retrofit an existing Search campaign: Go to Settings → AI Max and toggle on. Then set Search term matching, Text customization, Final URL expansion, and URL rules as needed. (Use this when you’re not running a controlled experiment.)

Quick pre-launch QA for Google Ads AI Max Campaign Experiments

• The campaign is Search-only, not on Shared budgets or a Portfolio bid strategy.
• No conflicting legacy features remain in this shell (older text customization, legacy brand inclusions/exclusions, ad-group location inclusion).
• If you keep Final URL expansion on, you’re OK with pinned RSA assets being ignored when a better URL is selected.
• Your team knows where to check Experiments → Summary and the AI Max reporting pages.

What exactly does AI Max change during Google Ads AI Max Campaign Experiments?

Google describes AI Max as “a one-click solution that brings the best of Google AI into your Search campaigns,” now available globally across Google Ads, Ads Editor, SA360, and the API.

The two pillars AI Max turns on in the trial arm

AI Max layers Search term matching and Asset optimization on top of your existing Search setup. During an experiment, these run only in the trial split unless you choose to apply them account-wide after the test.

Pillar	What it does in practice	Where you control it	What to watch in reports
Search term matching	Expands reach beyond your explicit keywords using broad + keyword-less matching informed by landing pages and intent signals.	Enable AI Max at campaign level, then toggle on/off per ad group as needed.	Search terms adds an AI Max match type so you can isolate incremental queries sourced by AI Max.
Asset optimization	Uses Text customization to assemble copy from your assets and site, and Final URL expansion to send users to the most relevant page.	Campaign → AI Max → Asset optimization; both toggles default on when you opt in.	Assets and Landing pages show AI-selected lines and a “Selected by” indicator for URLs.

Controls that keep experiments disciplined

AI Max adds a control surface so you can widen intent while protecting brand and UX. Use these before you launch the trial arm.

• Brand safety: Set brand inclusions/exclusions to steer which brand terms AI Max can match or must avoid. This helps keep competitor traffic clean during testing.
• Locations of interest: Guide keyword-less matches toward geo intent at the ad-group level for local or multi-market experiments.
• URL rules: Use URL exclusions at campaign level to block weak pages and URL inclusions at ad-group level to force landing on specific pages when you need precision.

Implementation tip (API): Campaign.ai_max_setting.enable_ai_max = true enables AI Max; set AdGroup.ai_max_ad_group_setting.disable_search_term_matching = true if an ad group should not use AI matching inside the same trial. This is useful for side-by-side ad-group reads.

Important behavior changes to understand (with caveats)

1. Search term matching is on by default once you enable AI Max. You can disable it per ad group to keep some ad groups keyword-only during the test. This is handy for layered hypotheses.
2. Text customization and Final URL expansion default on when you opt into AI Max. They work together. If you toggle off Text customization then Final URL expansion also turns off.
3. Pinned RSA assets can be overridden when Final URL expansion selects a different destination that requires different asset combinations. Plan copy guidelines with this in mind.
4. Reporting adds AI-specific visibility. You’ll see an AI Max match type in Search terms, plus aggregate rows like “AI Max expanded matches” in Keywords, and “Selected by” in Landing pages. These are the fastest places to quantify incremental coverage and routing quality.

What doesn’t change when you run Google Ads AI Max Campaign Experiments

• Bidding strategy & budgets: Your Smart Bidding goal and daily budgets remain intact inside the experiment shell. You only split traffic between control and trial.
• Campaign scope: AI Max is Search-only. It is not Performance Max and it will not push spend into YouTube, Display, or Shopping during the test.
• Experiment logistics: You still monitor the Experiments page, time-series charts, and status messages like any standard Search experiment. The twist is the AI Max toggles that live inside the trial.

A quick diagnostic table for common outcomes

Symptom in trial	Likely cause	Fast fix
CPA spikes with many new queries	Search term matching too permissive for your economics	Tighten with brand exclusions, add URL exclusions to remove weak pages, or disable matching for specific ad groups and re-test.
CTR dips on head terms	Text customization testing variants that don’t fit brand	Review Assets and remove misaligned autogenerated lines; keep high-performing pinned assets and consider pausing Final URL expansion temporarily.
Conversion rate varies by geo	Keyword-less matches drifting outside target intent	Add locations of interest at ad-group level; split geo-sensitive ad groups and re-run the trial.
Mixed landing pages hurt UX	Final URL expansion selecting broad pages	Add URL inclusions for priority pages or block weak pages with exclusions at campaign level.

Why these changes matter for measurement

Because AI Max runs inside your existing Search campaign, the only moving parts in the trial arm are matching and creative/URL automation. That tight scope helps you attribute lifts correctly. Use the AI Max match type in Search terms and the Landing pages “Selected by” column to separate incremental intent from creative routing effects before you call a winner. Then decide whether to auto-apply or roll out manually.

Success metrics, guardrails, and power analysis for Google Ads AI Max Campaign Experiments

When you run Google Ads AI Max Campaign Experiments, decide up front how you’ll judge success, how you’ll protect the test from noise, and how you’ll know you have enough data.

Pick the right primary KPI (and know where to read it)

Objective	Primary KPI for the experiment	Secondary diagnostics	Where to verify in Google Ads
Lead gen	Conversions, CPA	Lead quality, by geo or audience	Experiments → Summary for head-to-head lift; then Search terms (filter Match type = AI Max), Assets, Landing pages for root-cause context
Ecommerce	Conversion value, ROAS	AOV, product/URL mix	Same surfaces as above; use Landing pages → Selected by to see when Final URL expansion routed traffic.
Demand expansion	Incremental conversions, new query coverage	CTR, qualified query share	Search terms with AI Max match type and Keywords totals for “AI Max expanded matches.”

Guardrails that keep your AI Max experiment clean

1. Lock scope to Search. AI Max runs inside Search (it isn’t PMax), so judge it against Search KPIs and Search inventory only
2. Use the built-in experiment split. Keep control and trial in the same campaign to reduce setup drift; apply or discard after review
3. Rely on experiment sync (and avoid one-sided edits). If you change the base campaign during the test, Google can sync those changes to the trial; edits in the trial do not sync back. This preserves a fair comparison
4. Keep performance stable while testing. Google’s Campaign Guidance notes that consistent performance and sufficient volume improve “Experiment Power,” so avoid parallel changes that spike variability (targets, budgets, feed overhauls) during the window
5. Run long enough if results are undecided. If the Experiments UI shows In Progress/Undecided/Unavailable, Google recommends allowing 4-6 weeks to collect enough data before you judge
6. Use AI Max controls to prevent waste while you learn. Apply brand inclusions/exclusions, locations of interest, and URL include/exclude rules to keep reach disciplined during the trial

How Google determines significance (so you don’t guess)

Google publishes the math behind Ads experiments: the system buckets data, then uses jackknife resampling to estimate variance and runs two-tailed tests with 95% confidence intervals to call winners.

• What that means for you: call a win when the Experiments → Summary view shows a statistically significant improvement on your primary KPI. Don’t over-weight directional changes in secondary metrics; use them to explain why the primary moved (e.g., new AI Max queries or a different Selected by landing page).

Power analysis, the easy way (with Campaign Guidance)

Google’s Campaign Guidance surfaces an Experiment Power score with ranges Low (0-49%), Medium (50-79%), High (80-99%), plus specific steps to lift power – choose higher-volume campaigns, increase duration, and prefer a 50/50 split for faster, conclusive reads.

Quick checklist to raise power before you launch

• Pick a campaign with consistent volume (stable spend, steady conversions).
• Set a 50/50 traffic split unless risk requires a softer ramp.
• Extend duration if the UI flags low power; re-check the score after a few days.
• Avoid simultaneous structural changes that add noise (let the experiment isolate AI Max’s impact).

Where to look first (and why) during Google Ads AI Max Campaign Experiments

1. Experiments → Summary for the canonical head-to-head read and significance tag.
2. Search terms → filter Match type = AI Max to quantify incremental intent; use the combined search term → headlines → URL view to spot patterns.
3. Keywords → check the AI Max totals (“expanded matches” / “landing page matches”) to see total contribution beyond your list.
4. Assets and Landing pages → confirm automatically created lines and Selected by URLs, then tighten or expand controls accordingly.

Reporting for Google Ads AI Max Campaign Experiments – where and how to read results

You judge Google Ads AI Max Campaign Experiments in a few places. Start with the Experiments → Summary view for the head-to-head read. Then drill into Search terms, Keywords, Assets, and Landing pages to understand why performance moved. Google added AI-specific visibility like an AI Max match type, AI Max totals in Keywords, and a “Selected by” column for landing pages. Read each surface below and act on the insights right away.

Experiments → Summary (your canonical “who won?” view)

• What you get: A controlled control vs trial comparison with confidence intervals and a significance label on your primary KPI.
• How Google calls significance: Ads experiments use bucketed data + jackknife resampling and two-tailed 95% CIs to decide winners. You don’t need to calculate it yourself.
• Run time guidance: For trustworthy reads, Google recommends 4-6 weeks, and the first 7 days are excluded from the experiment’s result view to allow ramp-up.

Search terms report (with the new AI Max match type)

• What to look for: Filter Match type = AI Max to isolate incremental queries the trial unlocked. This helps you measure coverage gains and prune low-quality themes early.
• Practical read: Compare conversion rate and CPA on AI Max terms vs. your keyword-matched terms. Keep or negate themes based on economics.
• Where to click: Reports → Search terms → add “Match type” column → filter to “AI Max”.

Keywords report (AI Max totals)

• What to look for: Two helpful rows quantify AI’s impact beyond your list (for example, “AI Max expanded matches” and similar totals). Use them to size the lift even if long-tail terms are fragmented.
• Action: If totals add meaningful conversions at target CPA or ROAS, you likely have product-market signal worth rolling forward.

Assets report (text customization visibility)

• What to look for: Automatically created asset lines that AI assembled from your site and existing ads. Remove lines that violate brand style or compliance. Keep high-performers.
• Action: Pin only when compliance requires it. Otherwise let AI test variations during the experiment window.

Landing pages report (“Selected by” = AI choice)

• What to look for: A “Selected by” indicator shows when Final URL expansion routed a click to a more relevant page. Validate bounce rate and conversion paths on these URLs.
• Action: Add URL exclusions to block weak pages or URL inclusions at ad-group level when you need precision routing.

Reader’s map: where to read what (and what to do next)

Reporting surface	Key AI Max signal	Why it matters	Immediate action
Experiments → Summary	Lift on your primary KPI + significance	Decide apply vs discard with statistical backing	Apply or extend; document the decision window.
Search terms	AI Max match type	Quantifies incremental queries	Keep winners. Add negatives for waste.
Keywords	AI Max totals	Sums value from queries beyond your list	Size the lift. Judge if reach is worth the cost.
Assets	Auto-created lines	Shows which copy AI composed	Remove off-brand lines. Reinforce proven angles.
Landing pages	Selected by (AI-chosen URL)	Explains routing changes	Exclude weak URLs. Include priority pages.

How to isolate incremental queries in three clicks

1. Open Search terms.
2. Add Match type and filter to AI Max.
3. Sort by Conversions or Conv. value to find winners first.

Controls that keep AI Max disciplined during Google Ads AI Max Campaign Experiments

You want reach. You also want control. AI Max gives you both if you use the right dials. Set brand controls, locations of interest, and URL rules before launch. Then confirm changes with the AI Max reporting views.

Your AI Max control panel at a glance

Control	Level	Why you use it	Quick setup path	Key caveats
Brand settings (inclusions & exclusions)	Inclusions: Campaign or Ad group · Exclusions: Campaign	Allow or block specific brands to protect adjacency and keep tests clean	Settings → AI Max → Brands. Ad-group inclusions override campaign inclusions	Brand exclusions in Search upgrade into AI Max during rollout; manage them in the AI Max panel going forward.
Locations of interest	Ad group	Aim keywordless matches at high-intent metros or regions	Ad groups → AI Max → Locations of interest	Use with geo-sensitive offers. It guides expansion without broad geo sprawl.
URL exclusions	Campaign	Block weak or off-brand pages when Final URL expansion is on	Settings → AI Max → Final URL expansion → Add URL exclusions	Requires Text customization and Final URL expansion enabled.
URL inclusions	Ad group	Force landing on a curated set of URLs for precision flows	Ad groups → AI Max → Add URL inclusions	If you enable inclusions and Final URL expansion then Google can use your inclusions and predicted best URLs. Pinning may not be respected.
Search term matching	Campaign on → manage per Ad group	Expand beyond your list with broad and keywordless tech	Settings → AI Max → Search term matching, then per ad group	Turn it on at campaign level first. Toggle off in any ad group that must stay keyword-only.
Text customization (asset generation)	Campaign	Tailor headlines/descriptions to query intent	Settings → AI Max → Asset optimization	Enables Final URL expansion. Review and remove off-brand lines in the Asset report.
Final URL expansion	Campaign	Route clicks to the most relevant page on your domain	Settings → AI Max → Asset optimization	If on, pinned RSA assets may not be respected. Same when URL inclusions are enabled.

Tighten-the-loop playbook (use this during the experiment)

1. Seed guardrails first. Add brand exclusions/inclusions, set locations of interest, and establish URL exclusions/inclusions before launch. This keeps the trial disciplined from day one.
2. Let it learn. Google recommends waiting ~2 weeks before heavy edits like negatives. Give AI Max room to optimize.
3. Audit what changed.
   • Search terms → Match type = AI Max to isolate incremental queries.
   • Landing pages → Selected by to see AI-chosen URLs.
   • Asset report to prune auto-created lines.
     
4. Prune with purpose. Add negatives only when terms miss your goals. Avoid overusing negatives or you choke intent-based targeting benefits. Review search terms every 1-2 weeks.
5. Lock pins only if required. Final URL expansion can override pinned assets. Turn it off if you must enforce pins.

Scenario → control mapping

Scenario in the trial arm	What likely happened	Control to apply	Where to apply it
Non-brand CPA rises on competitor terms	Expansion hit rival brand queries	Add brand exclusions	Settings → AI Max → Brands (campaign)
Strong query. Poor landing path.	Selected by routed to a thin page	Add URL exclusions. Or set URL inclusions for priority pages	Settings → AI Max (campaign) and Ad groups → AI Max (ad group)
Geo quality varies	Keywordless matches drift by region	Use locations of interest for target metros	Ad groups → AI Max → Locations of interest
Off-brand ad lines appear	Text customization tried risky variants	Remove lines in Asset report. Add brand guidance internally	Reports → Assets
One ad group must stay exact/phrase only	You need a clean baseline	Disable Search term matching in that ad group	Ad groups → AI Max → Search term matching

How reporting proves your control strategy

• Search terms → AI Max match type quantifies the incremental queries your controls allowed or blocked.
• Landing pages → Selected by shows whether URL rules fixed routing.
• Asset report confirms that brand cleanup stuck.
  Use these three together after each change so you can connect configuration to business results.

Common pitfalls in Google Ads AI Max Campaign Experiments (and how to avoid them)

You can squeeze real gains from Google Ads AI Max Campaign Experiments if you dodge the usual traps. Below are the mistakes that blow up tests and the fixes that keep results clean and credible.

Launching with an ineligible setup

If your Search campaign uses Shared budgets, a Portfolio bid strategy, targets the Display Network, or still has legacy features toggled on, the built-in AI Max experiment will not run. Fix the shell first, then test.

Quick fix: Move the campaign to its own budget and bidding. Remove legacy text customization, brand inclusions/exclusions, and ad-group location inclusion settings from their old places. Then re-open the experiment flow.

Turning on Final URL expansion without guardrails

Final URL expansion can send users to a different page than your ad’s original final URL. It also requires Text customization and may ignore pinned RSA assets, which surprises teams that expect strict pinning.

Quick fix:

• Add URL exclusions at campaign level to block thin or off-brand pages.
• Use URL inclusions at ad-group level when you need precision landing.
• Enforce pins only if you must, or temporarily turn off Final URL expansion.

Skipping brand controls

Leaving brand settings blank invites competitor or off-brand matches during the trial. AI Max gives you brand inclusions/exclusions to keep reach disciplined.

Quick fix: Add brand exclusions at campaign level and brand inclusions at campaign or ad-group level before launch. Then audit the Search terms report with Match type = AI Max for stray themes.

Treating AI Max like Performance Max

AI Max is a Search-only upgrade. It does not shift budget to YouTube, Display, or Shopping. If you expect multi-channel scale from this test, you’ll misread the outcome.

Quick fix: Judge success on Search KPIs and Search inventory only. Use PMax for cross-channel scale if that is the goal.

Starving the test or stopping too soon

Low volume or short runs lead to inconclusive reads. Google’s experiments guidance recommends at least 4-6 weeks when results still show In progress/Undecided/Unavailable.

Quick fix: Prefer a 50/50 split for power. Extend duration if the verdict is undecided. Industry coverage notes Google’s built-in AI Max experiment often uses a 50/50 split by default.

Mixing structural changes during the trial

Editing targets, budgets, or creative across only one arm muddies your read. For custom experiments, Google’s experiment sync copies base-campaign edits one way into the trial, which prevents drift if you change the base. Edits in the trial do not sync back.

Quick fix: Make necessary changes in the base campaign only and let sync do its job in custom-experiment setups. For the built-in AI Max experiment, keep changes minimal during the window and judge in Experiments → Summary.

Letting AI Max roam without watching the right reports

Teams sometimes look only at headline KPIs and miss why they moved. AI Max adds:

• Match type = AI Max in Search terms
• AI Max summary rows in Keywords
• Selected by in Landing pages
• Asset performance for automatically created lines
  Use these to separate incremental queries from routing and copy effects.

Over-negativing in week one

Killing themes too early can choke the system before it learns. You still need to prune waste, yet you should base changes on AI Max segments in the Search terms view after sufficient data accrues.
Quick fix: Review Search terms → Match type = AI Max, then add surgical negatives. Pair with URL exclusions/inclusions instead of blanket negatives when landing paths are the problem.

Forgetting ad-group exceptions

Some ad groups must remain keyword-only for compliance or clean baselines. You can disable Search term matching per ad group while AI Max stays on elsewhere in the same trial.

Ignoring asset governance

Text customization can produce lines that do not fit your brand. That is by design since it tests variations. You must prune what you do not want.

Quick fix: Audit the Assets report during the test. Remove off-brand lines. Keep winners that lift CTR or CVR.

Misdiagnosing “lift” when it is just routing

A win might come from smarter Final URL expansion rather than better query coverage. The Landing pages report shows Selected by so you can confirm when AI chose a different destination.

Quick fix: Compare Search terms (AI Max) against Selected by URLs. Size how much of the lift came from new queries vs better landing paths. Then set URL inclusions/exclusions accordingly.

Expecting Editor/API parity for every toggle

Google warns that turning AI Max on or off may affect API requests that manage text customization and brand settings during rollout stages. Coordinate with your devs before switching.

Advanced plays with Google Ads AI Max Campaign Experiments (go beyond the basics)

You’ve covered setup and guardrails. Now squeeze more from your tests. The plays below stack AI Max for Search features with disciplined measurement so you scale reach without burning budget. Each move maps to why it works, how to set it up, and what to watch.

1) Split hypotheses by intent cluster inside one campaign

Run multiple ad groups that represent distinct intents (problem, solution, competitor, brand-neutral). Keep Search term matching on for exploration groups and off for strict baselines—all within the same AI Max trial.
How: Enable AI Max at campaign level, then toggle Search term matching per ad group.
Measure: In Search terms, filter Match type = AI Max to quantify incremental queries by intent cluster.

2) Use Locations of interest to guide keywordless reach

If quality varies by city or metro, steer AI to where intent is strongest without widening your geo targets.
How: Set Locations of interest at the ad-group level before launch.
Measure: Compare geo-level CPA/CVR for AI Max terms vs. keyword-matched terms.

3) Shape landing paths with URL inclusions/exclusions

Great queries can still flop if traffic lands on thin pages. Keep Final URL expansion, but fence it.
How: Add URL exclusions at campaign level to block weak pages; add URL inclusions at ad-group level for precision flows.
Measure: In Landing pages, use “Selected by” to confirm AI-chosen URLs and compare CVR/ROAS.

4) Pair AI Max trials with Campaign Guidance for power

Don’t guess on run length. Google’s Experiment Power score shows whether your test can reach significance and how to fix it (duration, split, volume).
How: Check Campaign Guidance before launch; favor a 50/50 split unless risk is high.
Measure: Track the power score mid-test; extend only when needed.

5) Automate multi-campaign rollouts via the Google Ads API

When a trial wins, flip on AI Max across sibling campaigns programmatically and keep settings consistent.
How: Use Campaign.ai_max_setting.enable_ai_max (API v21) to enable; manage controls through the AI Max settings.
Measure: Build a deployment log (when toggled, which controls) and monitor post-apply drift in the Experiments summary tiles.

6) Centralize reporting and governance in SA360 + Looker Studio

If your team lives in SA360, keep AI Max experiments consistent and pipe results into your BI layer.
How: Manage Search campaigns in Search Ads 360; connect SA360 to Looker Studio for executive dashboards.
Measure: Mirror Google Ads experiment KPIs and the AI Max search-term/landing-page views in your BI.

7) Prepare copy and routing for AI experiences in Search

Google is expanding where ads can appear within AI-driven results (e.g., AI Overviews and related surfaces). That ecosystem rewards clear CTAs and tight landing relevance.
How: Keep Text customization and Final URL expansion on during trials, then prune assets and URLs in reports.
Measure: Watch CTR/CVR deltas on AI-created assets and Selected by URLs; anticipate more AI-infused surfaces over 2025.

8) Test brand safety at scale with brand inclusions/exclusions

Expansion can pull in competitor or ambiguous brand intent. Use brand lists to control adjacency instead of blanket negatives.
How: Configure brand exclusions at campaign level; add brand inclusions at campaign or ad-group level when you want to allow certain brands.
Measure: In Search terms, filter Match type = AI Max and review brand patterns weekly.

9) Build “asset sprints” and retire losers quickly

Treat Text customization like a sprint. Keep what drives CTR/CVR; remove the rest.
How: Review Assets weekly during the test; remove off-brand or under-performing lines; leave high-performers unpinned so the system can remix them.
Measure: Asset-level CTR/CVR plus holistic CPA/ROAS; confirm changes in the next sprint’s results.

10) Use Editor/one-click experiments to scale your testing cadence

Because AI Max experiments run inside a campaign, you can run more tests with less ops effort.
How: Launch experiments from Google Ads (or Editor) with standard splits and consistent guardrails.
Measure: Track time-to-verdict across your portfolio; prioritize high-impact siblings next.

Blueprint A – Lead Gen lift with disciplined expansion

• Goal: Increase qualified conversions without raising blended CPA.
• Lever: Turn Search term matching ON in the trial; keep Asset optimization ON (default).
• Guardrails:
  • Campaign must be Search-only, no shared budgets/portfolio strategies.
  • Apply brand exclusions (campaign) and brand inclusions (campaign/ad group) before launch.
  • Add URL exclusions for thin pages; consider URL inclusions for high-intent forms.
• Setup (quick path): Campaigns → Experiments → Campaign features & settings → AI Max. Choose 50/50 split, 28 days. Review defaults (both pillars ON).
• Primary KPIs: Conversions and CPA (Experiments → Summary).
• Diagnostics to read:
  • Search terms → filter Match type = AI Max to isolate incremental queries.
  • Keywords → check the AI Max totals to quantify value beyond your keyword list.
  • Landing pages → verify Selected by URLs aren’t hurting form conversion.

Pass if: conversions ↑ and CPA within guardrail (e.g., ≤ +10%). Fail if: CPA > guardrail for 3 consecutive days after ramp. (Use Google’s “Experiment Power” in Campaign Guidance to confirm run length.)

Blueprint B – Ecommerce routing boost with Final URL expansion

• Goal: Lift conversion rate and ROAS by sending query traffic to the most relevant product or category page.
• Lever: Keep Text customization ON and Final URL expansion ON (default). Add URL exclusions for low-intent articles; optionally URL inclusions for high-margin categories.
• Caveat: Final URL expansion requires Text customization; when on, pinned RSA assets may not be respected if a different URL is selected. Plan copy accordingly.
• Primary KPIs: Conversion value and ROAS (Experiments → Summary). Secondary: bounce/session depth from analytics.
• Diagnostics to read: Landing pages → Selected by column; Assets report for auto-created lines to prune.
• Run: 28–42 days, 50/50 split for power.

Pass if: ROAS ↑ with stable or better CVR. Tuning: if CVR varies by city, add Locations of interest at the ad-group level to steer keywordless matches toward high-intent metros.

Blueprint C – Local services quality filter with Locations of interest

• Goal: Improve lead quality and close rate in target metros while expanding reach.
• Lever: Enable Search term matching, then add Locations of interest at ad-group level for your priority cities.
• Controls: Brand lists (to avoid competitor bloat), URL rules to keep users on service pages.
• KPIs: Qualified lead rate (CRM), CPA by metro (Ads).
• Readouts: Search terms (AI Max) by region; Landing pages → Selected by for local pages.
• Run: 28–35 days, ramp with 70/30 if your market is seasonally fragile, then move to 50/50 to reach significance.

Pass if: metro-level CPA holds target while qualified rate ↑. Fail if: competitor brand leakage; fix with brand exclusions.

Blueprint D – B2B SaaS non-brand scale without cannibalizing brand

• Goal: Add non-brand conversions at target tCPA/tROAS without inflating branded cost.
• Lever: Search term matching ON for exploration ad groups; OFF for brand/control groups inside the same trial.
• Controls:
  • Brand exclusions at campaign level; inclusions for allow-listed partners.
  • URL inclusions for product and pricing pages; exclude docs/blog if they depress CVR.
• KPIs: Non-brand CPA/ROAS, incremental conversions from AI Max rows.
• Readouts: Search terms filtered to AI Max (non-brand themes), Keywords AI Max totals, Assets to remove off-brand lines.

Pass if: incremental non-brand conversions appear at or under target CPA. Rollout: apply to sibling markets and keep brand lists synced.

Blueprint E – “Broad vs AI Max” baseline check

• Goal: Decide whether campaign-level Broad match alone suffices or whether AI Max adds measurable incremental value.
• Design: Keep a pure Broad match Search campaign as baseline. In a sister campaign, run the AI Max experiment (with Search term matching + Asset optimization). Compare each campaign’s controlled experiment result to its own control arm to avoid cross-contamination.
• What to watch: If AI Max wins, you should see AI Max match-type terms and AI-selected landing pages driving the delta, not just broad interpretation of existing keywords.
• Decision: If the AI Max trial significantly beats the broad baseline’s control (per Ads’ CIs), standardize on AI Max in that cluster.

Pro tips before you hit “Create”

• Favor a 50/50 split for faster significance unless risk demands a gentler ramp. (Industry coverage notes 50/50 as the common pattern in AI Max experiments.)
• If results are Undecided, extend to 4-6 weeks and re-check power; don’t over-edit in week one.
• Keep expectations grounded; AI Max is Search-only. If you need cross-channel lift, that’s a Performance Max project, not an AI Max test.

Implementation checklist for Google Ads AI Max Campaign Experiments

You want a clean test that reaches significance and scales safely. Use this end-to-end checklist to stand up Google Ads AI Max Campaign Experiments, govern them during the run, then apply the winner with confidence.

Pre-flight: eligibility and setup sanity

• Confirm the campaign is Search-only. Do not target the Display Network.
• Remove Shared budgets and Portfolio bid strategies from the shell.
• Clear legacy toggles that now live under AI Max: old text customization/ACA, legacy brand inclusions or exclusions, and ad-group location inclusion.
• Open Campaigns → Experiments → Campaign features & settings → AI Max to start the built-in split. This runs the control and the trial inside one campaign which speeds learning and reduces configuration drift.

Why this matters: Google’s built-in AI Max experiment reduces duplication and lets you apply the winner back to the base with one click after review.

Configure the trial arm (the AI Max toggles)

• Search term matching: Enable at the campaign level first. Then decide per ad group which ones explore and which ones stay keyword-only.
• Asset optimization: Keep Text customization ON and decide on Final URL expansion. Remember that Final URL expansion requires Text customization and can ignore pinned RSA assets when routing to a different URL. Plan copy rules accordingly.
• Brand controls: Add brand exclusions at campaign level and use brand inclusions at campaign or ad-group level to allow specific brands.
• Geo guidance: Add Locations of interest at ad-group level to steer keywordless matches toward intent-rich metros.
• URL rules: Use URL exclusions at campaign level to block weak pages. Use URL inclusions at ad-group level when you need precision landing.

Pick a split and a timeline that reaches significance

• Default to a 50/50 split for faster reads and higher experiment power unless risk forces a softer ramp.
• Plan 4-6 weeks if the Experiments UI shows “In progress / Undecided / Unavailable”. Google’s experiments guidance and methodology back longer runs when power is low.
• Use Campaign Guidance to check the Experiment Power score and get tips to improve it.

FAQ – Google Ads AI Max Campaign Experiments

What exactly is AI Max for Search and how do “Google Ads AI Max Campaign Experiments” fit?

AI Max is a feature layer for Search campaigns that adds two pillars – Search term matching and Asset optimization – plus new reporting. Experiments let you split traffic inside the same Search campaign to compare AI Max on vs off then apply the winner.

Is AI Max a new campaign type like Performance Max?

No. AI Max is not a new campaign type. It is an upgrade to a Search campaign. Performance Max is a separate multi-channel type.

Where do I enable AI Max and start the built-in experiment?

Open Campaigns → Experiments → Campaign features & settings → AI Max then pick your traffic split and duration.

What are the main eligibility blockers?

The built-in experiment will not start if the Search campaign uses Shared budgets, a Portfolio bid strategy, or targets the Display Network. Legacy toggles like old text customization or brand settings can also block setup.

What’s new in reporting when I run Google Ads AI Max Campaign Experiments?

You get:

• Search terms with a new Match type = AI Max and a combined view of search term → headlines → URL
• Keywords with AI Max summary rows
• Assets and Landing pages with a Selected by indicator when AI chose the URL

How long should I run the experiment and what split should I use?

Google recommends at least 4-6 weeks for many tests. A 50/50 split speeds significance and is the common pattern noted in trade coverage.

How does Google call a winner in the Experiments summary?

Google buckets data then uses jackknife resampling to estimate variance and shows two-tailed confidence intervals in product.

Can I keep some ad groups keyword-only during the trial?

Yes. Turn Search term matching on at the campaign level then disable it per ad group that must remain strict.

Will AI Max change my landing pages automatically?

If you enable Final URL expansion then AI can route to a more relevant page and the Landing pages report will show Selected by for those clicks.

Does Final URL expansion require Text customization and can pins be ignored?

Yes. Final URL expansion requires Text customization. When Final URL expansion chooses a different destination your pinned RSA assets may not be respected.

How do brand controls work in AI Max experiments?

Use brand exclusions at the campaign level to block brands and brand inclusions at the campaign or ad-group level to allow specific brands.

What are URL inclusions and exclusions and when should I use them?

• URL exclusions (campaign) block weak or off-brand pages during Final URL expansion
• URL inclusions (ad group) force landing on a curated set of URLs

Is AI Max live in Google Ads Editor, SA360, and the API?

Yes. Google announced global availability across Google Ads, Google Ads Editor, Search Ads 360, and the Google Ads API.

How does AI Max build ad copy and what should I watch?

Text customization can compose assets from your site and existing ads then assemble them into a candidate ad. You should review Assets in reporting and remove lines that don’t fit your brand.

Does AI Max affect channels beyond Search like Performance Max does?

No. AI Max runs inside Search. Performance Max spans multiple channels.

Can I manage AI Max experiments and toggles via API?

Yes. The Google Ads API supports AI Max for Search campaigns and documents setup steps for developers.

How do I read incremental query coverage quickly?

Open Search terms then add Match type and filter to AI Max. This isolates queries unlocked by AI Max during the trial.

Do ads appear in new AI-driven Search experiences and does that affect experiments?

Google is expanding ads across AI experiences in Search which increases the importance of relevant copy and routing in your Search stack. You should monitor Assets and Landing pages during experiments since AI surfaces can amplify mismatches.

Looking to push Google Ads AI Max Campaign Experiments even further? Pair this playbook with three deep dives: tighten inventory quality and audit reach with the Google Ads Search Partner Network placement report guide, steer query expansion toward high-intent metros using Google Ads AI Max Locations of Interest, and stress-test creative at scale with Google Ads Demand Gen asset experiments.