Judge an AI Max experiment by the cost and quality of the leads your team can actually serve, alongside the conversion data reported in Google Ads. Before raising a budget, define a qualified lead, connect that definition to your records, and allow time for inquiries to become sales opportunities.
A local service business can receive more forms while its calendar gets no busier. Some people live outside the service area. Others need a service the business does not provide. A few are duplicates. Each submission looks like activity; each still costs someone time to review.
First, check the current migration dates
Google’s AI Max migration announcement, updated June 11, says Dynamic Search Ads auto-upgrades begin in February 2027. Automatically Created Assets and campaign-level broad match upgrades start in September 2026. The older DSA September date still appears in some coverage. Use the dated update and your account’s actual notices when planning changes.
Google also reports an average 7% increase in conversions or conversion value at similar CPA or ROAS for the full AI Max feature set versus search-term matching alone. Its footnote identifies Google internal 2026 data for non-retail advertisers. That is a vendor aggregate, not a forecast for your business or evidence of a 7% increase in qualified inquiries.
Define “qualified” before looking at the winner
Write criteria your sales team can use consistently. For a service business, that might mean the requested service is offered, the location is covered, the contact is genuine, and the proposed timing is workable. Avoid changing the definition halfway through a test because one campaign’s numbers look disappointing.
Google distinguishes qualified leads and converted leads as later outcomes in a lead process. Your own stages should reflect real business decisions. A form submission, a qualified inquiry, a booked consultation, and a paying customer should not be treated as interchangeable evidence.
A worked example: lower form cost, higher opportunity cost
The following numbers are hypothetical, not an Opslumo client result or a Google benchmark. Suppose two comparable test groups each spend $1,000:
- Group A produces 50 forms and 20 qualified leads: $20 per form and $50 per qualified lead.
- Group B produces 80 forms and 16 qualified leads: $12.50 per form and $62.50 per qualified lead.
Group B’s cost per form is 37.5% lower, but its cost per qualified lead is 25% higher. It also creates more submissions to review. Neither result tells us profit: we still need to know which opportunities become customers, their value, and the cost of serving them.
The calculation is simple: advertising spend divided by the number of leads meeting the agreed definition. Keep the denominator visible in reports. Ten qualified leads and one hundred qualified leads do not carry the same uncertainty even when their average costs match.
Design a test your team can interpret
Google’s August 20 testing announcement describes expanded experiments and planning tools. Availability and account setup still need checking. A forecast is a planning input; it is not observed customer demand.
- Confirm the conversion setup. Check duplicates, stage definitions, and whether the reporting includes primary or secondary actions.
- Document the change. Record which AI Max features, geographic controls, landing pages, and budgets differ.
- Keep sales handling comparable. Similar response times and qualification rules help avoid confusing a follow-up problem with an advertising result.
- Wait for meaningful outcomes. Use your actual sales cycle and available volume. There is no universal seven-day answer for low-volume businesses.
- Review the search-to-page match. Check whether inquiries came from people seeking an offer that the destination page clearly supports.
If importing later outcomes, use the platform’s supported measurement setup and your organization’s privacy requirements. Do not put customer details into public spreadsheets or treat hashed identifiers as permission to ignore consent requirements.
Decide what happens after the test
Set your review conditions in advance: acceptable qualified-lead cost, service capacity, evidence quality, and the minimum maturity of outcomes. An inconclusive test is a valid result. With very few leads, extend observation or make a smaller operational adjustment rather than announcing a winner.
If suitable inquiries are arriving but going unanswered, fix the handoff first. Our CRM follow-up guide covers the next step between a captured lead and a useful conversation.
Source and scope note: Official Google announcements and help documentation were reviewed on October 2, 2026. The calculations above are illustrative. This article does not promise a campaign result or recommend a specific advertising budget.