Name the setup before you touch the settings
A contractor can run ordinary Search, activate AI Max within Search, use standard Performance Max or manage a specialized pay-per-lead campaign migrated from LSA. A screenshot saying "Performance Max" is not enough to determine which instructions apply. Confirm the campaign goals, placements and migration notice in the actual account.
Google describes AI Max as a suite of search-term matching and asset-optimization features. It can extend matching beyond existing keywords using broad and keywordless technology. Standard PMax is a separate cross-channel campaign. The specialized local pay-per-lead version keeps LSA-style local placements and billing. It does not open ordinary cross-channel inventory.
That split matters in 2026. Video and Display advice can fit standard PMax. Keep that advice off a specialized pay-per-lead campaign. A migrated LSA campaign can sit inside Google Ads without ordinary Search keyword management.
| Setup | Primary role | Important boundary |
|---|---|---|
| Search without AI Max | Search campaign with selected keyword strategy | Still requires measurement and query review |
| AI Max for Search | Expanded matching and asset optimization | Search feature suite. Performance Max is a different product |
| Standard Performance Max | Cross-channel conversion campaign | Not LSA pay-per-lead billing |
| PMax with pay-per-lead goals | Specialized local lead campaign | Search and Maps only; keywordless |
What the 2026 LSA migration changes
Google's transition page says selected U.S. Home and storefront advertisers began moving in August 2026, with broader groups later in 2026 and remaining categories or regions in 2027. Your specific account notice controls timing. Do not assume every contractor has migrated, especially categories subject to advanced verification.
The documentation says settings and lead history transfer. Historical campaign performance reports do not come with them. Save whatever performance evidence you can pull before the move. The notes also describe weekly budgets converting to daily averages, deprecation of manual lead bidding, and a unified campaign-level target CPA where applicable. Open the account and check the settings you actually have. Write those settings down. The next person should be able to read them without remembering the old dashboard.
Pay-per-lead campaigns continue on Search and Maps and remain keywordless. Google Verified status carries over for already verified advertisers. Ordinary Performance Max keeps its own billing. The pay-per-lead campaign is the one that charges for leads. Keep the product distinction visible in reports so an office owner does not approve a different buying model by accident.
For AI Max, audit the pages AI is allowed to choose
Google says enabling AI Max turns on search-term matching and asset-optimization settings. Final URL expansion requires text customization and may choose a more relevant page on your domain. For a contractor, that can help intent alignment, or send a paid visitor to an informational article instead of a service appointment page.
Inspect URL exclusions and any ad-group URL inclusions supported in the account. Consider careers, unrelated resources, outdated offers and pages for work you do not currently sell. Review generated text for unsupported arrival promises, financing language or licensing claims. The website itself becomes an advertising input, so stale copy is not harmless.
Check tracking with an expanded landing page. Google's help page warns that incompatible tracking templates can create broken destinations. Also note its warning that pinned responsive-search assets may not be respected when a more relevant URL is chosen. If wording is mandatory for your business, do not assume a pin alone preserves it under expansion.
- ☐ List service pages eligible for paid traffic
- ☐ Exclude careers and irrelevant content where appropriate
- ☐ Remove expired offers and unsupported claims from source pages
- ☐ Test tracking on a dynamically selected destination
- ☐ Review brand inclusions and exclusions at the correct level
- ☐ Confirm service-area controls and actual capacity
- ☐ Check generated assets and the matched query together
A clear lead definition comes before the automation
Automation follows the events you give it. If every short call, repeated inquiry and spam form counts as equal success, more optimization can produce more of the wrong activity. Define qualified demand and completed work separately, connect reliable CRM outcomes and deduplicate repeat events before evaluating a new feature.
Keep secondary observations separate from primary bidding goals. A brochure download may be useful context without being an appropriate acquisition objective for emergency service. Assign values only when the business can defend them. Do not give a model-generated score the same status as a confirmed sale.
Compare conversion definitions across test and control. A result is uninterpretable if the experiment uses qualified leads while the baseline still counts raw forms. Track work type and coverage too: a small repair inquiry and a replacement estimate may have different economics even when both are legitimate leads.
Worked example: a cheaper form can still cost more per job
Illustrative experiment: a baseline spends $2,000 and produces forty raw inquiries, twenty qualified opportunities and eight completed jobs. A test spends the same amount and produces sixty inquiries, eighteen qualified opportunities and six completed jobs. Raw inquiry cost falls from $50 to about $33.33.
In this worksheet, qualified-opportunity cost rises from $100 to about $111.11, and completed-job acquisition cost rises from $250 to about $333.33. The figures are hypothetical. Use them to read the worksheet, and keep them out of any performance claim. They show why a lower dashboard cost per conversion can hide a worse business result when conversion definitions are too shallow.
Before declaring a winner, account for sales lag and sample size. Planned roofing jobs can close after the reporting window. Check whether staff handled both groups similarly and whether capacity changed. A handful of jobs should prompt investigation. One test does not prove that a Google product always works, or that it never works.
For standard PMax, use the reporting that now exists
Google's current channel performance report exposes campaign-level channel contribution, costs, conversions and diagnostics for supported campaigns. Use it to see where standard PMax is serving and whether the conversion mix makes sense. Do not repeat outdated claims that no channel reporting exists.
Channel reporting shows where the spend went. It does not give you a separate budget for each channel. Google says advertisers cannot directly allocate the standard PMax budget across channels in that report. Use the supported controls, assets, exclusions and goals, and inspect lead quality after changes. If the business needs a tightly scoped Search objective, compare a suitable Search setup. A cross-channel campaign does not run like a Search campaign.
Current lead-generation guidance also discusses search-term review and negative keywords for applicable inventory. Validate availability and scope in the account. Do not copy these controls onto migrated pay-per-lead campaigns. Those campaigns use service types and areas, and ordinary keyword management belongs to a different product.
Run one bounded test with stop conditions
Write the hypothesis before changing settings: for example, whether AI Max can find additional qualified replacement-estimate demand without increasing unsupported-service inquiries. Choose a stable baseline, a defined spend ceiling and the actual business outcome. Avoid simultaneously changing offer, landing page, bidding goal and automation if you want an interpretable result.
In the first days, check for broken URLs, the wrong service area, claims you are not allowed to make, and junk intake. Stop and fix those before you judge the campaign. A clean first day is too early to call a winner. Allow the relevant conversion lag and review the uncertainty. If the campaign is already budget constrained, Google warns that AI Max may not be effective, so do not treat expansion as the fix for every account.
Keep a change log and a rollback plan. Document which controls were enabled, which pages were eligible, and which conversion actions drove bidding. If you cannot explain the test or roll it back, leave the headline metric out of the decision.
Which decision should a small contractor make first?
If an LSA migration notice is present, preserve historical reporting and validate the transition before starting a separate automation experiment. If Search measurement is unreliable, repair it before enabling broader matching. If landing pages make promises the office cannot fulfill, fix them before allowing more automated asset generation.
When those fundamentals are sound, choose the test that fits the objective and available evidence. AI Max can be evaluated as a Search expansion layer. Standard PMax is a cross-channel decision. Pay-per-lead local campaigns remain their own operational path. None is a substitute for answering calls or having capacity for the work advertised.
Name the campaign in the account before you turn on another product with AI in the name. Learn its controls and billing, and measure whether it produces work you can deliver at a profit.
Sources
- Google Ads: how AI Max for Search campaigns works, reviewed November 2026
- Google Ads: LSA transition to PMax with pay-per-lead goals and 2026 to 2027 timeline
- Google Ads: about specialized pay-per-lead goals
- Google Ads: current channel performance report
- Google Ads: PMax lead-generation best practices