Microsoft Advertising’s 27 August 2026 AI Max launch gives Search advertisers a new automation surface for matching, text, assets and landing pages. It is similar in shape to Google’s AI Max language, but it is not the same product. The account, default state, reporting and migration checks belong to Microsoft Advertising.
What Microsoft AI Max changes
Microsoft describes search-term matching beyond the existing keyword list, text customization from assets and site content, and final URL expansion. It also describes reporting for an AI-optimized match label, assets and the search-term/landing-page relationship. Those reports are important because the automation changes not only copy but the relationship between query, asset and destination.
Microsoft says existing negative keywords, bidding, budget, conversion and targeting controls continue to be respected. DSA remains supported. Treat these as documented product behaviour to verify in the actual campaign, not as permission to skip a field-by-field read-back.
Default state and migration
Microsoft says AI Max is enabled by default for new Search campaigns and can be turned off. A new-campaign checklist should therefore include an explicit enable/disable decision, owner, date and reason. Do not let a default become an undocumented strategy.
Google Import is described as preserving settings. Before and after import, export or screenshot the campaign’s negatives, budget, bid strategy, conversion goals, audiences, locations, language, schedules, assets, final URLs and tracking templates. Reconcile every difference. A successful import message is not proof that a destination, measurement tag or exclusion behaves identically.
Matching, copy and final URLs
Expanded matching can find demand that is not written as an exact keyword. Review the query and conversion-quality report for intent, brand safety, geography, exclusions and lead quality. Text customization can produce relevant copy, but it still needs legal, offer, brand and landing-page review. Keep pinned or required messaging rules documented.
Final URL expansion can help match a query to a relevant destination, but it can also expose weak templates, stale pages, forms with broken parameters or pages outside the intended campaign scope. Test redirects, canonical URLs, tracking parameters, consent, forms, page speed and claim consistency before expanding.
A guarded test plan
Choose one campaign or a clearly bounded campaign group. Freeze objective, budget ceiling, conversion definition, audience and date window. Define a control or holdout where the platform supports it. Set guardrails for qualified conversion rate, cost, disallowed terms, landing-page errors and customer-service impact.
Read back the AI-optimized match label, search terms, assets and landing pages on a fixed cadence. Compare platform conversions with CRM or revenue quality. Preserve a rollback instruction that names the approver and the fields to restore. If the actual account differs from the announcement, record the discrepancy rather than generalizing from documentation.
How to read Microsoft’s performance claim
Microsoft reports 44 experiments conducted between June and August 2026, a spend-weighted average conversion increase of 13.6% and a lower confidence bound of 8.2% for ten statistically significant experiments. These are Microsoft’s internal results. They do not establish a causal effect for every industry, budget, conversion definition or import path, and they are not a target to copy into a forecast.
Run the team’s own measurement: incremental qualified outcomes, cost per qualified action, revenue, search-term quality, landing-page defects, asset review time and budget variance. Keep the vendor claim in the evidence section and label your result separately.
Default-state and import-control audit
For a new Search campaign, add an explicit AI Max decision to the build checklist even if Microsoft says the feature is enabled by default. Record whether it is on or off, who decided, the date, campaign objective and reason. For an existing campaign, do not infer the state from the new-campaign default. Read the actual campaign setting and capture the account and market.
For Google Import, export the source and destination settings before the import, then read back the destination after synchronisation. Compare negative keywords and terms, bid strategy, budgets, conversion actions, audiences, locations, language, schedules, assets, final URLs, tracking templates and any brand or term exclusions. “Settings preserved” is a product claim to verify, not a substitute for a migration diff. Keep the source export available for rollback.
Matching, copy, URL and reporting audit
Review a representative query set before and after AI Max. Include brand, non-brand, long-tail, competitor, location, ambiguous and known-negative terms. For each query, record the AI-optimised match label where shown, search term, asset or text variant, destination, conversion status and reviewer decision. A broader match is acceptable only when the query serves the intended reader job and meets the account’s quality rules.
Text customization should be checked against required claims, offers, legal language, tone, character limits and the landing-page content. Keep a list of pinned or mandatory messages and confirm what the platform actually served. For final URL expansion, test the homepage, product/service page, location page, redirected page and parameterised URL. Verify canonical, redirects, tracking, consent, forms, page speed and offer continuity.
Use the reporting labels to investigate, not decorate a post-campaign story. If an asset or landing page cannot be traced to a query and date, mark the evidence gap. Keep a read-back table with campaign, setting, query, asset, URL, date, outcome and action.
DSA boundary and campaign architecture
Microsoft says Dynamic Search Ads remain supported. DSA is a separate campaign approach with its own targeting and page-selection logic; do not describe continued DSA support as proof that every DSA behaviour is identical to AI Max. Keep campaign type, feed/page rules and exclusions explicit when comparing results.
Avoid changing DSA, AI Max, budget, bid strategy and landing pages in one unlabelled migration. If the business needs a combined change, document it as a multi-variable rollout and do not attribute the result to AI Max alone. A platform comparison with Google AI Max can be useful for terminology, but the controls, reporting and default state must be read from each platform.
CRM-qualified outcome measurement
Define the outcome before launch: qualified lead, booked appointment, completed purchase, revenue or another business event. Preserve the platform click or conversion identifier where permitted, deduplicate repeated leads, and record disqualified reasons, sales response time and revenue status. Compare AI Max and the control on qualified rate and cost per qualified outcome, not only raw platform conversions.
Capture the date of any CRM, form, consent, call-tracking or offline-conversion change. If the platform reports a conversion increase while qualification or revenue falls, pause and investigate. Microsoft’s internal experiment result is not a target, confidence interval for this account or evidence that a migration will pay back.
Rollback and approval
Set a maximum budget, a landing-page error threshold, a disallowed-query rule, a qualified-rate floor and a named approver. Keep the pre-change export and the post-change read-back. If matching expands into irrelevant or unsafe demand, a URL breaks, a tracking template returns an error, spend paces beyond the cap or CRM quality deteriorates, pause AI Max or restore the prior campaign state according to the approved plan.
Record the pause time, affected campaigns, fields restored, remaining ads and unresolved evidence. Do not delete the experiment data. A rollback is a controlled decision, not proof that the feature is universally bad; a positive result is not proof that the same setting belongs in every campaign.
Context for DMT readers
The DMT AI for Google Ads guide is a Google-focused comparison owner; this article keeps Microsoft AI Max’s defaults and reports separate. The target-bidding changes guide and language-targeting guide help readers hold adjacent assumptions stable, while the Google Ads API v25.1 guide is migration context rather than Microsoft implementation evidence.
Bottom line
Microsoft AI Max deserves a platform-specific test plan because it changes matching, copy and URLs while adding new reporting. Inspect the default state, read back every imported control, test destinations and review qualified outcomes. Keep Microsoft and Google AI Max separate, and treat the reported 13.6% conversion increase as internal vendor evidence rather than a promise.
Credits
- Central facts: Microsoft Advertising primary launch post; accessed 29 August 2026.
- Supporting context: Search Engine Land and PPC discussion; terminology only.
- Internal-link verification: refreshed authenticated DMT WordPress inventory.
- Visual credit: DMT-authored migration and control diagrams.
Related DMT guides
Sources
- Microsoft Advertising: AI Max for Search is now available — primary source accessed 29 August 2026.
