DSAs disappear: AI Max exits beta and the web becomes key

After more than 10 years managing Google Ads campaigns, We have experienced many changes: the arrival of Smart Bidding, the transition to responsive ads (RSA), the adoption of Automated Shopping, and its subsequent disappearance, the emergence of Performance Max, and many other movements that have been transforming the platform.

Each of these changes has been reducing manual control and increasing the weight of automation in decision making.

And with the announcement made by Google on April 15, the same thing happened again. But this time we are not looking at a new configuration or an incremental improvement: it is a change that forces us, once again, to rethinking how campaigns are built and optimized.

As of September 2026, campaigns with Dynamic Search Ads (DSA) will be automatically migrated to AI Max. If we don't act first, Google will do it for us (once again).

And with this move, something that until now was secondary becomes critical: the way in which the client's website is built.

A bit of history: DSA already read your website

Before entering what are AI Max campaigns, It is worth remembering what DSAs were and why from the beginning they were directly related to the way in which a website is built.

Dynamic Search Ads worked by crawling the advertiser's site to automatically generate ad titles. and select the most relevant landing pages for each search.

In practice, Google was analyzing the web in much the same way as it does for the organic positioningIt interpreted its content and, from there, decided which searches to activate and which pages to respond with.

What did this imply on a day-to-day basis?

What the quality of the website already marked the campaign's performance.

If the structure was confusing, if the pages did not have a clear hierarchy, if the URLs were not descriptive or if the content was poor, the DSAs performed worse. Not because of campaign configuration, but because the base from which Google started was not good.

This resulted in less relevant searches, less qualified traffic and, in many cases, the need to pause or limit the campaign.

On the other hand, when the web was well worked, the performance changed completelyWe segmented by URL categories, excluded sections that were not of interest and controlled which pages should or should not be included.

What is AI Max and what makes it different

AI Max is not simply «DSA with AI». It is a system that combines three functionalities that previously existed separately:

• Search term matchingAI identifies relevant searches beyond the keywords you have configured, based on your web content and real-time intent signals.
• Text customizationAd text generation: it generates ad texts adapted to each search, drawing on the content of your pages.
• Final URL expansionThe site's landing page: autonomously chooses which page of the site is the most relevant for each specific search, even if it is not the URL that you had indicated as the destination.

These three features have something in common: all three depend on the AI understanding your website well.

If the Google bot reads your website and finds pages with vague titles, non-existent header structure, duplicate content or URLs without logic, it will make poor decisions. It will send users to the wrong pages. It will generate generic titles (and customers will call us to complain) and display ads on searches irrelevant to the business.

Five SEO aspects that a GAd manager can audit

It's not about becoming a technical SEO overnight. But it is about knowing what signals we are giving Google's AI, and making sure they are the right ones. These are the critical points.

1. URL architecture and site structure

The way a website is organized is the first level of signal the AI receives. A URL such as: /services/insurance/civil-liability-autonomous-employers

clearly explains what the page is about.

In contrast, a parameter-based URL does not provide useful information.

In addition, elements such as the sitemap or internal linking help to define which pages are relevant within the site.

For AI Max to work, we need the site to have:

• Clean, hierarchical and descriptive URLs, without unnecessary parameters or numeric IDs as unique identifiers.
• An updated XML sitemap, which acts as an explicit sitemap. If a page is not in the sitemap, it has a good chance of not being crawled or considered by AI Max.
• Consistent internal linkingImportant pages should receive links from other pages on the site. An «orphaned» page, without any internal link pointing to it, is practically invisible to the crawler and therefore to the AI.

2. H1, H2 and H3 heading tags: the semantic skeleton of each page.

Headings are the way a page tells Google (and AI Max) what it's about, at what level of detail, and what it's about. how the information is organized.

The H1 is the main title of the page. There should be only one per page, and it should clearly describe the main topic. It sounds obvious, but in practice it is one of the most frequent problems: pages with generic H1 («Welcome», «Our services», «Home»), pages without H1, or pages where the H1 is the name of the company and not the subject of the content.

For AI Max, this has a direct effect. When the functionality of text customization generates the title of an ad based on the content of a page, part of the headings. An H1 like «Liability Insurance for Self-Employed in Madrid» gives you a rich signal. An H1 like «More information» is useless.

The H2 and H3 add semantic structure: they tell the AI which subtopics are covered by the page., The more rich and well-structured this hierarchy is, the better the automatically generated ads will be. The richer and more well-structured this hierarchy is, the better the automatically generated ads will be.

3. Title and meta description: the tags that AI reads first

The