Google Ads AI Max: A Shopify Audit After September 1

Google began automatically upgrading eligible Search campaigns to AI Max on September 1, 2026. The rollout is phased, and Google expects eligible upgrades to finish by the end of September. This is not a new campaign type. It is an optimization layer that can change query matching, generated ad text, and, when enabled, landing-page selection inside existing Search campaigns. The practical issue for Shopify teams is not whether AI Max is universally good or bad. It is whether the account still targets commercially useful demand, presents accurate offers, sends shoppers to valid pages, and optimizes against reliable purchase value.

Google's updated migration announcement limits the September trigger to Search campaigns using automatically created assets, now called text customization, or the campaign-level broad match setting. A Shopify store that used neither feature is not automatically included merely because it runs Search ads.

What Is Actually Changing

Google says the automatic migration will mirror the legacy setup rather than enable every AI Max feature in every campaign. The published mapping is:

Important clarification: Manually enabling AI Max is different from being auto-migrated from one legacy setting. Google's standard opt-in flow enables search term matching, text customization, and Final URL expansion by default, although each can be adjusted. The audit must record what is active in the live account, not infer it from the AI Max label alone.

The Six-Step Shopify Audit

1. Confirm Which Campaigns Migrated and When

Start with scope. Review account notifications, campaign Settings, the AI Max columns, and Change history. Record the migration timestamp, the previous trigger, the responsible owner, and whether the change was automatic or manual.

For agencies or teams managing campaigns through code, Google Ads API v25.1 added migration date-time fields for automatically created assets and campaign-level broad match. These fields can support an account-wide inventory, but the UI remains the place to inspect the enabled subfeatures and controls.

Record for each affected campaign:

• migration time and legacy trigger
• search term matching status at campaign and ad-group level
• text customization and Final URL expansion status
• brand inclusions, brand exclusions, location controls, and negative keywords
• URL inclusions, URL exclusions, and tracking template
• bid strategy, budget, primary conversion action, and value settings

2. Preserve a Comparable Baseline

Export the last clean pre-migration period before changing settings. Use a period long enough to cover the store's normal buying cycle, and preserve the same weekday mix for later comparison. Save campaign and ad-group settings, keywords and match types, negatives, search terms, ads and assets, landing pages, spend, clicks, conversions, conversion value, and bid targets.

If the campaign has already migrated, do not try to recreate a theoretical baseline by memory. Use the final complete pre-change period, mark the migration date, and annotate promotions, stockouts, price changes, consent changes, or tracking releases that could distort the comparison. A before-and-after chart without these notes can assign normal commercial variation to AI Max.

3. Audit Query Expansion Before You Judge ROAS

Search term matching can use broad match, keywordless matching, ad assets, and landing-page content to find additional queries. Google's AI Max reporting exposes an AI Max match type, a source column, and a combined view of the search term, headline, and landing page. Review that combined journey, not the search term in isolation.

Review in this order:

• highest-spend new queries, because poor expansion becomes expensive there first
• queries that convert but produce low-margin, cancelled, or returned orders
• brand queries that may be taking credit from existing demand
• competitor, informational, support, employment, and policy queries that do not match the campaign's job
• location-sensitive queries where the store cannot deliver, offer pickup, or honor the advertised terms

Use controls at the level where the problem occurs. Search term matching can be disabled for a specific ad group when AI Max remains active at campaign level. Brand exclusions are campaign-level controls; brand inclusions can be set at campaign or ad-group level. Locations of interest can steer geographic intent. Negative keywords still have a role when a query is clearly irrelevant or unsafe.

4. Govern Generated Text and Landing Pages Together

Text customization can generate headlines and descriptions from the domain, landing pages, existing ads, keywords, and other assets. Google tells advertisers to monitor generated text for accuracy. For a Shopify store, that means checking more than tone. Generated copy must not invent a discount, delivery promise, product attribute, warranty, stock position, or payment option.

If Final URL expansion is active, Google can replace the supplied final URL with another relevant page from the same domain. Google's Final URL documentation also states that pinned responsive-search-ad assets are not respected when a different URL is selected. Review the URL and the generated message as one unit.

Exclude or repair pages that create a broken commercial journey, such as:

• sold-out or discontinued products with no suitable alternative
• policy, account, support, cart, or internal-search pages that are not valid ad destinations
• market or language versions showing the wrong price, currency, delivery area, or legal terms
• thin collection pages whose filters or pagination produce unstable URLs
• pages where a subscription, COD, pickup, trust, or add-to-cart component is missing or conflicts with the ad claim

The storefront is part of the campaign input. Progus's guide to Shopify product data for AI shopping provides a useful companion check for titles, variants, availability, images, and policies. For offer language, make sure delivery, return, and payment claims follow the same facts presented in the store's customer-facing buying information.

Finally, test tracking templates. Google warns that static or malformed templates can produce the wrong destination or a 404 when Final URL expansion substitutes a dynamic page. Use the supported {lpurl} patterns and test representative expanded URLs before allowing broader traffic.

5. Prove Measurement From Ad Click to Shopify Order

AI Max can only optimize toward the conversion signals it receives. Shopify's Google & YouTube channel can connect a Google Ads account for conversion measurement even when a merchant does not use Merchant Center product syncing. Confirm that the intended purchase action is active, receives the correct order value and currency, and is not duplicated by another pixel or tag.

This check is especially important for stores that recently changed Thank You or Order Status tracking. The Progus post-purchase tracking audit explains how app pixels, custom pixels, consent, and duplicate purchase events can affect reporting. Do not increase AI Max spend while the purchase signal is missing, duplicated, or using the wrong value.

Run controlled journeys covering:

• a product page and a collection page selected by the campaign
• mobile and desktop checkout
• the store's main markets, currencies, and languages
• an unavailable variant or out-of-stock product
• subscription, COD, pickup, or other conditional purchase paths used by the store
• consented and non-consented analytics journeys where applicable

For every test, verify the query, served headline, selected URL, page content, checkout completion, purchase event count, order value, currency, and Shopify order. Shopify's order conversion summary can help inspect referrals and UTM parameters for individual orders, but Shopify notes that details can take up to 48 hours to appear. Differences between platforms are normal because attribution models and reporting times differ; unexplained breaks and duplicates are not.

6. Decide With Commerce Outcomes, Not Platform Lift Alone

Google Ads can show whether AI Max generated additional traffic and conversion value. It cannot decide whether those orders were commercially useful for the merchant. Build the review around four evidence groups:

• Reach: AI Max spend, new-query spend, search term source, brand versus non-brand mix, and landing pages selected by AI Max.
• Journey quality: irrelevant-query spend, generated-text errors, URL mismatches, 404s, unavailable products, and checkout or app failures.
• Commercial value: Shopify orders, revenue, gross profit after discounts and variable costs, new-customer share, cancellations, and returns.
• Measurement health: missing or duplicate purchases, value and currency errors, consent effects, attribution shifts, and reporting lag.

Where an eligible experiment is available, use it instead of relying only on a before-and-after comparison. Google's AI Max experiment splits traffic between a control with AI Max off and a treatment with it on. Google advises allowing an inconclusive experiment to run for at least four to six weeks, and longer if there is not enough data. New multi-campaign budget and ROI tests are also rolling out during September, so availability can vary by account.

Keep, Narrow, or Disable

End each review with one explicit decision:

• Keep and scale: when AI Max adds profitable, relevant demand and the generated message, destination, and measurement remain reliable.
• Narrow: when value is concentrated in certain ad groups or URLs. Use ad-group search-term controls, negatives, brand controls, URL rules, or disable text customization or Final URL expansion while retaining useful parts of AI Max.
• Disable and repair: when conversion tracking is untrusted, generated claims are repeatedly wrong, destinations break, or incremental traffic cannot meet the store's margin and customer-quality thresholds.

Google's setup documentation allows advertisers to disable AI Max or adjust individual features. The right configuration is the smallest set of automated capabilities that produces verified commercial value without weakening offer accuracy or measurement.

Final Thoughts

The September migration is not a reason to reject automation, and it is not permission to accept every default. AI Max connects campaign settings more tightly to the words and pages on a Shopify store. That makes query quality, storefront accuracy, and conversion measurement part of the same operating system.

Record what changed, inspect the journeys Google creates, and judge them against Shopify orders and profit. When the evidence is clean, scale deliberately. When it is not, narrow the feature that caused the problem instead of treating AI Max as one indivisible switch.