Google expands AI Mode ads to Australia

NEWS / AUSTRALIAN EXPANSION

1. AI Mode ads reach Australia through existing campaigns

Announced 28 October · Help checked 30 October 2026

What changed

Google’s 28 October announcement names Australia among ten new markets for ads in AI Mode. It separately describes the launch of ads in follow-up questions as a US development. That US launch should not be read as confirmation of the same feature in Australia.

The linked, undated Google Ads help page lists Australia and English for ads within AI Mode. It says eligible Search, Shopping and Performance Max campaigns can participate automatically through relevant auctions without a separate campaign or budget. Exact and phrase match can participate on directly commercial queries; AI-powered targeting can also support contextual matching.

The help page says AI-surface performance is included in standard campaign and search-term reporting. Advertisers cannot target only AI Mode or separately opt out of it. Existing campaign controls still apply. Google also says buying ads does not influence organic AI answers.

Who this affects

Australian businesses already using eligible Google campaigns, including service providers whose ads may match relevant commercial searches. Actual serving in a particular account is not established here. Businesses that do not advertise can ignore campaign changes; this is not a reason to buy ads to influence organic answers.

Our analysis: what it means for your business

Hypothetical example: a Melbourne air-conditioning installer already advertises for replacement systems. A new place where ads may appear does not change its service boundary, available installation capacity or minimum viable job. Before paying for broader demand, the owner still needs those limits reflected in the campaign and the page an enquirer reaches. This is an operating example, not an account result.

Our recommendation is to start with the existing business measure: qualified enquiries and booked work relative to total spend. If the account mainly counts page visits or unqualified form submissions, an increase in reported conversions would not answer the owner’s question. Ask the operator which conversion actions represent useful demand and how staff distinguish enquiries they can serve from those they cannot.

Aggregate reporting also limits attribution. A change in total clicks or costs after the announcement cannot, by itself, prove that AI Mode caused it. Keep a dated record of campaign changes and review enquiry quality alongside spend. A country appearing in the availability list is evidence of a supported market, not evidence that any particular ad has appeared there.

Do not broaden targeting or increase a budget simply to pursue a new placement. First check whether the current destination page states the service, area, exclusions and next step clearly. Any later campaign experiment should have a defined business question and an agreed spend limit. Google’s announcement cites US product-list click data; it does not establish Australian service bookings or a return on advertising spend.

What to do next

1. Review qualified conversion definitions and actual enquiry quality with the existing operator.
2. Check service-area exclusions, ad promises and destination pages before considering a targeting or budget change.
3. Record any authorised experiment separately from the rollout date; judge total business outcomes without claiming AI Mode attribution that the evidence cannot support.

NEWS / NEW CLIENT PORTAL

2. Portal service booking depends on the calendar setup

Announced 29 October · Checked 30 October 2026

What changed

HighLevel says customers can now discover and book services within the new Client Portal. Customer details can be prefilled, with single or multiple service bookings depending on the setup. Meetings and service bookings appear together under upcoming and past appointments.

Availability is controlled at two levels: enable client-portal booking on the chosen service booking page, and enable Appointments in the portal’s app permissions. Eligible rescheduling and cancellation follow the calendar settings. The announcement describes the new Client Portal; it does not label this release as a Labs feature.

Who this affects

Appointment-based service businesses already using the relevant portal and service calendars. Growthcenter users need their operator to verify the specific account and settings. A business with a working booking link and no customer demand for a portal can safely leave that process alone.

Our analysis: what it means for your business

Hypothetical example: a mobile detailing business offers several packages but has only one vehicle and a limited travel area. Putting all packages in a portal may make them easier to find, but it does not prove that the calendar prevents incompatible bookings or allows enough travel time. The useful check is whether a customer can choose a service the team can actually fulfil.

Our recommendation is to expose only the services intended for self-booking, then follow one authorised sample booking through the calendar, confirmation, reminders and staff handoff. Check which customer details are carried through and which still need to be collected. A booking visible to the customer must also be visible to the person responsible for delivering it.

Rescheduling needs its own check. If a customer changes an eligible appointment, the business needs the old slot, new slot and relevant staff notifications to remain consistent. Cancellation rules should match what the customer is told before booking. These are proposed acceptance checks, not faults observed in this release.

This service-booking update is distinct from yesterday’s report on the Labs portal builder and custom CRM records. The vendor’s convenience claims do not establish fewer calls, lower no-show rates or more bookings in a particular business. Any benefit depends on customers using the portal and staff trusting the complete booking process.

What to do next

1. Confirm new-portal access and which service pages are intended for customer self-booking.
2. Use an authorised sample booking to check capacity, confirmation, reminders and staff ownership.
3. Test an eligible reschedule and cancellation before inviting customers to depend on those controls.

NEWS / LABS DEPENDENCY

3. An AI Builder test can send a real message

Announced 29 October · Checked 30 October 2026

What changed

HighLevel announced workflow testing inside AI Builder chat. Its instructions require enabling The New Test Workflow in Labs. Operators can select a contact or use events, inspect step inputs and outputs, and read a summary of completed or skipped steps.

The vendor explicitly says the test executes real actions: tags can be added, notes written and messages sent. Some actions are skipped. Testing remains optional before saving and publishing. This release is an in-chat testing workflow, distinct from the private-beta workflow API covered in the previous edition.

Who this affects

Businesses whose operator uses AI Builder for enquiry, reminder or follow-up workflows. Growthcenter account and Labs access remain unverified. Owners who do not use this builder need no setting change.

Our analysis: what it means for your business

Hypothetical example: an operator drafts a missed-call follow-up and chooses a real customer as the test contact. If the selected path sends an SMS, the word “test” does not make that message private to the operator. The source describes real effects, so the test plan must identify the recipient and connected actions before execution.

Our recommendation is to use contact details controlled by the business for authorised tests, inspect the data first and review every destination that could be reached. An owned phone number does not by itself make an unrelated webhook, staff alert or external record update harmless. Keep the test narrow enough that the operator understands and can reconcile its effects.

Read the skipped-step list as carefully as the success summary. A skipped action has not proved that part of the process works. Where a booking or message outcome matters, record the actual result at the receiving end and separately investigate any untested path. A plain-language summary can help navigation, but it does not replace evidence of delivery.

No workflow was run and no customer message was sent for this article. The announcement establishes a testing interface and its stated effects, not an assurance that AI-generated workflows are correct or that one test covers every customer situation.

What to do next

1. Verify the Labs requirement and use only authorised, controlled test contact data.
2. Inspect message recipients and other connected effects before starting a run.
3. Record actual outcomes and skipped actions separately before deciding whether the workflow is ready for use.