Google Ads AI Max: What to Audit Before the 2026 Upgrade

Google Ads is moving several familiar Search campaign settings into AI Max, and the transition is close enough that advertisers should stop treating it as a distant product announcement. Google says eligible campaigns using automatically created assets or the campaign-level broad match setting will begin moving to AI Max in September 2026. Dynamic Search Ads have more time, with their transition now scheduled to begin in February 2027.

But the real decision is not simply whether to switch a toggle. AI Max can influence which searches a campaign enters, which copy appears, and which page receives the click. That makes campaign setup, conversion data, website content, tracking, and account controls one connected system. A clean system can give the automation useful room. A messy one can spread the mess.

Product note: This article reflects public Google Ads documentation available on August 14, 2026. Rollout dates, settings, and feature availability can change. Check the current account interface and Google’s documentation before changing a live campaign.

Key takeaways

• AI Max is an optimization layer for Search campaigns, not a separate campaign type.
• Automatically created assets and campaign-level broad match remain scheduled to begin upgrading in September 2026, while the Dynamic Search Ads transition is scheduled for February 2027.
• The highest-risk inputs are weak conversion goals, irrelevant eligible URLs, stale website claims, incompatible tracking templates, and missing campaign controls.
• Negative keywords still apply, and AI Max adds brand, location, URL inclusion, and URL exclusion controls that should be reviewed before a test.
• A controlled experiment should be judged by qualified business outcomes, not by a larger conversion count alone.

What is AI Max for Search campaigns?

Google describes AI Max as a continuous optimization layer for existing Search campaigns. It brings three main systems together: search term matching, text customization, and final URL expansion. Search term matching can use broad match, ad assets, and landing-page signals to find relevant queries. Text customization can create ad copy from your ads, assets, and website. Final URL expansion can select a different, query-relevant page from your domain.

Turning on AI Max enables text customization and final URL expansion by default, although those campaign-level settings can be turned off individually. Search term matching is controlled at the ad-group level. That matters because an advertiser can use AI Max without giving every feature the same amount of freedom.

It is still a Search campaign. Budgets, bidding, conversion goals, ads, exclusions, location choices, and landing pages remain the advertiser’s responsibility. AI Max changes how those inputs can be combined and expanded. It does not make weak inputs harmless.

September 2026 and February 2027 are different deadlines

Google’s original April announcement placed Dynamic Search Ads, automatically created assets, and campaign-level broad match on the same September 2026 transition path. The company updated that announcement on June 11. The DSA transition was postponed, but the other two settings stayed on the September schedule.

Do not use September as a deadline for every DSA decision. And do not use the DSA delay as a reason to ignore automatically created assets or campaign-level broad match. They are separate transitions. Record which campaigns use each legacy setting, then build a review calendar around the setting that actually applies.

Understand the three systems you are turning on

These systems are connected. Final URL expansion requires text customization so the generated message can match the selected page. If final URL expansion or URL inclusions choose another page, pinned responsive search ad assets may not be used. A campaign built around legally required pinned wording needs a different test plan from a campaign with flexible copy.

Audit conversion data before expanding reach

Automation learns from the outcomes an account defines. In Google Ads, primary conversion actions can be used for bidding, while secondary actions are generally kept for observation. If a newsletter signup, page view, form start, and qualified sales lead all carry the same weight, the campaign may find more easy actions without finding more valuable customers.

Picture a service business that records 40 form starts, 12 completed forms, and three qualified inquiries in a month. If bidding sees the form start as the main success event, a rise from 40 to 55 may look good even when completed forms and qualified inquiries stay flat. That is not an AI Max defect. It is a measurement definition doing exactly what it was told to do.

• Confirm that every primary conversion represents a business outcome worth bidding toward.
• Remove duplicate tags and imported goals that count the same action twice.
• Count a completed form or confirmation event, not a click on the submit button.
• Set call rules that separate meaningful conversations from short or unanswered calls.
• Use conversion values that reflect real differences instead of assigning an arbitrary high value to every lead.
• When possible, return qualified-lead or closed-sale information so campaign reports do not end at the first form submission.

No tracking setup can guarantee better campaign performance. But clean goals make the test interpretable. If the account cannot agree on what a valuable conversion is, expanding the search and landing-page space will create more uncertainty, not less.

Treat the website as part of the campaign inventory

Final URL expansion can send a click to a relevant page on the advertiser’s domain instead of the final URL originally entered in the ad. That turns the eligible website into campaign inventory. A page that was harmless as a low-traffic resource can become expensive when paid clicks start arriving there.

Start with a URL export, not a meeting about whether AI is good or bad. Review every page that could plausibly match the campaign theme. Open it on a phone. Check the offer, location, price language, form, phone number, load behavior, and next step. Make sure the page says what the business can actually deliver today.

• Exclude careers, login, privacy, support, thank-you, and other pages that are not valid first destinations for paid traffic.
• Exclude expired promotions, unavailable services, sold-out products, and old event pages.
• Review blog posts and guides individually. Some support a buying decision, while others answer a question without offering a useful next step.
• Use URL inclusions or page-feed labels when an ad group should stay within a defined section of the site.
• Keep a written reason for every broad exclusion so useful pages are not blocked by habit.

A blanket exclusion of every blog page can be as careless as allowing every page. The right question is whether the page matches the search, represents the offer accurately, and gives the visitor a sensible path forward.

Website copy can become ad copy

Text customization can draw marketing messages from existing ads, assets, and URLs, then create assets that fit the query and selected page. That means a forgotten sentence on a landing page can influence a live ad. Old pricing, unsupported superlatives, expired offers, and service-area mistakes deserve the same review as the responsive search ads already in the account.

Regulated advertisers need extra care. If a required disclosure, qualification, or legal phrase must always appear in a pinned position, Google says pinning may not be respected when final URL expansion is on or URL inclusions are provided. If pinning is necessary, turn off final URL expansion and do not provide URL inclusions. Then verify the setup and get appropriate compliance review before testing.

After launch, use the combined Search terms view to review searches, headlines, and URLs together, then use the asset and landing-page reports for their performance detail. Do not wait for a complaint to inspect generated language. A weekly review during the first month is a reasonable operating habit, even when performance looks stable.

Set boundaries before asking for more reach

AI Max does not erase advertiser controls. Google says negative keywords are respected. Brand inclusions and exclusions can define which brands a campaign should or should not be associated with. Locations of interest can guide geographical intent in keywordless matches. URL inclusions and exclusions shape the landing-page set.

Review these controls together. A negative keyword can block the wrong query but cannot repair a misleading landing page. A URL exclusion can keep traffic away from a careers page but cannot correct a conversion goal that counts every button click. Brand controls can protect a boundary, but they do not decide whether a lead was profitable.

Also check budget and bidding. Google warns that AI Max will not be effective when a campaign is limited by budget, and its documentation says AI Max search term matching will not work with Manual CPC. That does not mean every advertiser should raise a budget or change bidding. It means the test design must acknowledge the limits of the current setup.

Test tracking templates with dynamic landing pages

Google specifically warns that final URL expansion can conflict with some tracking templates and produce broken dynamic landing pages, including 404 errors. Static tracking URLs and unusual uses of the {lpurl} parameter are two documented causes. This is a technical pre-launch check, not something to investigate after paid traffic disappears.

• Test the account, campaign, ad-group, ad, and keyword tracking settings that can affect the final click URL.
• Use several eligible landing pages, not just the homepage or the URL already attached to one ad.
• Confirm that parameters persist without creating redirects, malformed URLs, or duplicate question marks.
• Check mobile and desktop destinations and test any third-party measurement redirect.
• Verify that analytics and CRM fields still receive the campaign details needed for lead-quality reporting.

If a tracking provider owns the template, ask it to confirm compatibility in writing. Keep the answer with the launch record. A passing test on one URL is useful, but a dynamic system needs a representative set of pages.

Run an experiment with a written baseline

Google offers a one-click AI Max experiment that splits traffic 50/50 between the original campaign and the AI Max treatment. The cleanest test begins with a written hypothesis, stable conversion definitions, a fixed observation period, and a short list of decision metrics. Avoid rebuilding the landing pages, changing the bid target, and redefining conversions in the middle of the same experiment.

Here is a hypothetical HVAC example. The campaign includes repair, installation, and maintenance pages. Careers, privacy, an expired rebate page, and broad informational posts are excluded. Calls longer than 60 seconds and completed appointment forms are primary actions. Form starts remain secondary. Employment terms and do-it-yourself searches are added as negatives after a search-term review, not from a copied list that might block valid demand.

The business then compares qualified calls, booked appointments, cost per qualified inquiry, conversion value, search-term relevance, and selected landing pages. If AI Max produces more form fills but fewer booked jobs, the larger conversion count is not a win. If it finds valuable searches at an acceptable cost without creating copy or destination problems, the account has evidence for a wider test.

Who should test now, and who should wait?

Waiting does not have to mean rejecting AI Max. It can mean spending two weeks fixing measurement, page eligibility, or ownership so the eventual test answers a real question. Moving early without a readable baseline is faster only on the calendar.

A 12-point AI Max pre-launch checklist

1. Inventory every campaign using automatically created assets, campaign-level broad match, or Dynamic Search Ads.
2. Record the current conversion goals, bid strategy, budget status, search terms, landing pages, and qualified results.
3. Separate primary bidding actions from secondary observation events.
4. Remove duplicate conversion tags and test every form and tracked call path.
5. Review negative keywords, brand controls, locations of interest, and ad-group boundaries.
6. Export candidate website URLs and exclude those that should never be eligible for paid traffic.
7. Read every eligible landing page for stale offers, unsupported claims, wrong locations, and weak next steps.
8. Decide whether final URL expansion, URL inclusions, or page-feed rules fit each ad group.
9. Confirm whether pinned assets are mandatory and adjust the asset-optimization plan accordingly.
10. Test tracking templates and parameters across a representative set of dynamic landing pages.
11. Write the experiment hypothesis, observation window, success metrics, and stop conditions before launch.
12. Assign weekly owners for search terms, generated assets, selected landing pages, lead quality, and business outcomes.

Review the first 30 days as a system

During the first week, check for obvious destination, tracking, query, and copy problems. In the second week, compare qualified outcomes and review which landing pages AI Max selected. In weeks three and four, look for patterns that have enough volume to justify an exclusion, inclusion, page improvement, or wider test.

Do not grade the campaign on one headline or one strange search term. And do not excuse a repeated pattern as a learning phase. A single bad match may be noise. A group of expensive irrelevant matches, broken URLs, misleading assets, or low-quality leads is an operating issue that needs action.

Our view is simple: AI Max can widen the useful search space, but it also connects parts of the marketing system that teams often manage separately. The paid search account, website, conversion tracking, and sales-quality feedback need to agree on what the business wants. The toggle comes last.

If the transition is approaching and your team cannot tell which campaigns are affected, which pages are eligible, or which conversions guide bidding, Ardoz Digital can audit the account and website together. The review can produce a cleaner test plan and clearer controls. It cannot promise lower costs, more conversions, or better lead quality.

Ardoz Digital is not affiliated with or endorsed by Google. Google Ads and related product names are used only to identify the products discussed.