AI Max widens the way Google Ads Search campaigns find queries, create messages and choose landing pages. That can open up opportunities, but it also calls for controls. If you sell B2B services, the decision is not simply to switch on "more AI": it is to check which searches, ads and pages represent your offer well.
Google Ads requires an advertising budget. This guide explains its controls; it is not a free promotion route nor a recommendation to launch campaigns.
What changes with AI Max in Search campaigns?
Imagine an architecture studio advertising full home renovations. Its campaigns run on a set of terms it knows well. AI Max can widen the matching to other searches, adapt certain texts and send traffic to pages that the system considers relevant. The studio, however, does not want enquiries for small repairs that it does not do.
That is the real problem: advertising coverage is not the same as commercial fit. The technology can explore signals that were not in the original keyword list, but the advertiser remains responsible for defining the service, the exclusions and the result it considers valuable.
According to Google's documentation, AI Max for Search campaigns combines search term matching, text customisation and final URL expansion. They do not all have the same effect. The first influences the queries for which an ad can appear. The second adapts messages. The third can widen the landing pages that are used.
The risk is not that "AI does whatever it wants" by definition. It is setting up a campaign without understanding its permissions, its limits and its conversion signals. An old page for a service that has been withdrawn, for example, can create expectations that the sales team cannot meet if it enters the advertising circuit.
Not all accounts receive the same features at the same time. Before changing anything, check the options actually available in your interface and the current Google Ads documentation.
Source: Official help for setting up AI Max. Accessed: 09/10/2026.
Which updates did Google announce in September 2026?
On 23 September 2026 Google announced the expansion of the AI Brief closed beta to several languages, including Spanish. AI Brief lets you guide AI Max with context about the company, the audience and the message. "Closed beta" means precisely that you should not assume your account has access.
Google also previewed a new reporting view that links the search that triggered an ad, the creative assets shown and the final page. The company said it would share more details and availability later. So it would be wrong to say that this report is already live for every advertiser.
There is another change worth reading carefully. Google initially communicated a migration of legacy features to AI Max. In an update on 11 June 2026 it clarified that the retirement and automatic migration of Dynamic Search Ads would start in February 2027, while certain Automatically Created Assets settings and campaign-level broad match would follow the September 2026 schedule.
This difference avoids a common headline mistake: not all the old features changed date in the same way. Before reviewing a dynamic ad, identify the campaign type and the specific setting it uses.
The relevance for a small business is immediate even if it does not use the beta features. If you hand Google Ads to a provider, you need to know which changes can affect traffic, which controls remain and how decisions are documented.
Sources: AI Max updates, 23/09/2026 and Migration update, 11/06/2026. Accessed: 09/10/2026.
Which controls should you review before switching it on?
Start with the offer, not the switch. A lead-generation campaign for an employment law firm should not show the same message to someone looking for ongoing advice for companies as to someone who needs a one-off personal consultation, if the firm only handles the first.
It is worth reviewing five decisions:
- Search coverage. Which services do you offer, and which requests do you not want to attract?
- Landing pages. Which URLs should stay out of automatic expansion?
- Messages. Which commercial claims need review before they are published?
- Geography and brand. Which markets and proper names make sense for your activity?
- Conversion. Which event represents a useful commercial opportunity, and which is just an interaction?
Google documents URL exclusions at campaign level and other control options. Exclusions do not replace reviewing the site: if you have out-of-date pages, the problem is also editorial and technical. A campaign should not be the mechanism for permanently hiding an inconsistent website.
Also avoid using a simple increase in clicks as proof of quality. A click can come from a query that is too broad. A form can contain a request that does not fit. A call may not turn into an opportunity. Each signal represents a different stage.
If you work in regulated sectors, promises and generated messages need additional controls. The technical ability to customise a text does not allow you to attribute results, credentials or conditions that are not substantiated.
How do you tell more form submissions from better opportunities?
The useful metric depends on what happens after the form. Imagine a company that installs industrial air-conditioning systems. It receives requests from private homes and from factories. If it only works with industrial installations, counting both contacts as equivalent can steer the campaign towards the wrong audience.
Agree with sales a simple, checkable classification: contact received, valid contact, qualified opportunity, proposal sent and deal closed. Not every company needs those five stages, but it does need a shared definition of what it wants to achieve.
Then check whether the data can travel between the form and the CRM without losing the source. The goal is not for the algorithm to know every personal detail about the customer. It is to have proportionate signals, with an appropriate legal basis, that help analyse the quality of lead generation.
In September 2026 Google published guidance on data strength in lead generation campaigns and talked about initiatives to connect commercial information and bidding. That does not mean every feature mentioned is available on every account, nor that they should be implemented without a privacy review.
An indicator that is often overlooked is the reason for discarding a lead. If half of the requests a team receives do not match the advertised service, knowing why they are discarded gives you more insight than celebrating a low cost per form. You do not need to publish invented figures: it is enough to record the real reasons.
Source: Google, lead generation campaigns and commercial data, 18/09/2026. Accessed: 09/10/2026.
How do you test AI Max without losing visibility?
A responsible test starts with a specific hypothesis: "we want to check whether widening coverage brings us enquiries from companies in the sector that we had not anticipated". It does not start with "let's switch everything on and see".
Document the starting state: affected campaigns, settings, exclusions, pages, goals and conversion events. Then review the controls available for running an experiment. On 20 August 2026 Google announced AI Max testing tools and options to evaluate changes to budget and goals, with gradual rollouts.
Before comparing, define what a better result would mean. For a consultancy, the number of relevant commercial conversations may matter more than the volume of forms. For a shop, the relevant figure may be attributable sales, taking returns and margins into account. There is no universal answer.
Keep a log of changes. If you change the website, the ads, the targeting and the sales response process at the same time, you will not easily be able to attribute the result to AI Max. That methodological limit matters more than any screenshot with a rising curve.
Nor should you turn a test into a promise of return. Campaigns depend on the market, the competition, the quality of the offer and the commercial follow-up. Google publishes aggregate results for some features, but they are not a guarantee for your company.
Source: Google, AI Max testing tools, 20/08/2026. Accessed: 09/10/2026.
How have we approached it at meiika?
At meiika we understand Growth & MarTech as the connection between message, data and sales process. Before proposing campaign changes, it matters to understand which kind of customer the business wants to attract and how each request is handled.
We also work on websites because a confusing landing page can damage the experience even when the targeting is right. We do not attribute to meiika any campaign results that are not documented, nor do we promise that AI Max will automatically improve an account.