AI Max URL Exclusions: Google Explains Why Your Rules May Be Leaking Traffic

If you run AI Max on a Search campaign and rely on AI Max URL exclusions to keep paid traffic off your blog, careers page or last year's offers, check your rules this week. Google has published a Help page, About URL exclusions in AI Max for Search campaigns, that explains how "URL contains" rules actually read your text. The matching is far stricter than most advertisers assume, and a rule that looks like it covers a whole section of your site may be blocking one folder and nothing else.
What Google has documented
The page sits in Google's AI Max help section and carries no publication date. Search industry coverage of it appeared over the last few days. It does not say whether the behaviour is new or has simply been written down for the first time. Either way, it is now the official explanation of how your exclusions are applied, and that is what matters.
The basics first. URL exclusions only work when both Text customisation and Final URL expansion are switched on. They are a campaign-level control, and according to Google's AI Max setup guide, any exclusions you have set are ignored if AI Max is turned off. You can exclude pages in three ways:
- Exact URLs
- Custom labels from a page feed
- Custom "URL contains" rules
The first two do what you would expect. The third is where the trouble is.
How "URL contains" actually reads your rule
It matches whole words, not fragments. Google's own example is a rule for "auto", which will not exclude a URL containing "automation". If you typed a short stem hoping it would catch every longer word that starts with it, it will not.
Hyphens, full stops and slashes split a URL into words. Underscores do not. Google treats something like promo_page as one unbroken token, so a rule for "promo" will not touch it.
Plurals and numbers break the match. An exclusion for "FAQ" blocks /FAQ/page, but not /faqs or /FAQ11.
A space means AND, anywhere in the URL. Put a space in your rule and Google splits it into separate words that can appear anywhere in the URL, in any order. Google recommends not using spaces at all, and we agree.
What that looks like on a real site
Applying those rules to typical URLs (worth testing against your own site before you rely on it):
- A rule for "blog" blocks /blog/meta-ads-tips, but not /blogs/, /blog2025/ or /blog_archive/.
- A rule for "sale" blocks /black-friday-sale/, but not /sales/ or /summer_sale/.
- A rule for "careers" blocks /careers.html, because the full stop counts as a separator.
- A rule for "trade account", with the space, blocks any URL containing both words anywhere. That could include /account/trade-prices/, a page you may well want live.
Most modern platforms use hyphens in URLs, which helps. Older sites, custom builds and anything migrated from another platform often mix in underscores, plural folder names and dated slugs. Those are exactly the cases where a tidy-looking exclusion list quietly leaks.
Why this matters more than it looks
Final URL expansion is switched on by default when you turn on AI Max. That means Google, not you, decides which page each click lands on. Your exclusion list is the main thing standing between paid traffic and your job adverts, your privacy policy or an offer that ended in March.
A click that lands on the wrong page is not just wasted. It also feeds Smart Bidding a signal that the search term does not convert, which can push the campaign away from queries that would have worked with the right page. We wrote about why landing pages decide so much of your conversion rate in PPC landing pages, and this is the automated version of the same problem.
If you upgraded to AI Max in September, read what the AI Max upgrade changed alongside this. Exclusions you set back then were written before Google explained how they work.
Dynamic Search Ads users: fix your rules before the upgrade
If you still run Dynamic Search Ads, this has a deadline attached. Google's help page says that once a DSA campaign is upgraded to AI Max, legacy rules such as "page title contains" and "page content contains" become read-only. You can delete them, but you cannot edit them or add new ones.
The Google Ads Developer Blog sets the timetable: the ability to create DSAs is removed in January 2027, and any active DSA campaigns are automatically upgraded from February 2027. Right now is the only point at which those legacy rules can still be changed. Review them before you upgrade, or before Google does it for you.
What we would do this week
- List every Search campaign with AI Max on, and note whether Final URL expansion is enabled. If you did not choose it, it probably is.
- Export your real URLs from your sitemap or a crawl, then read every rule against them. Look for short stems, plural folders, numbered slugs and underscores.
- Replace stems with the exact word, and add plural or variant versions as separate rules.
- Remove spaces unless you genuinely want the AND behaviour.
- Use exact URLs or page feed labels for anything that must never receive paid traffic: checkout, account, careers, legal pages. Do not trust a pattern with those.
- Check the landing pages AI Max is choosing, weekly, in the same session you review search terms and negative keywords.
Our honest view: for a lead generation site with a few dozen pages, Final URL expansion rarely earns its keep. You already know which page should answer each search, so switch it off and send traffic where you have built it to convert. For a retailer with a large catalogue, it can find useful product pages you did not think to target, but only if the exclusion list is watertight. Audit it before Black Friday traffic arrives, not after.
Need a second pair of eyes?
AI Max settings are one of the first things we check when we take on a Google Ads account, alongside the landing pages the traffic actually reaches through our conversion rate optimisation work. If you would like us to review your exclusions before peak season, get in touch.

