AI Max Final URL Expansion on Hospital Sites: Sending Patients to the Right Doctor Page

This is for hospital marketing teams and website owners who have AI Max switched on, or are considering it, and want to know where paid traffic actually lands. Final URL expansion lets Google choose a landing page other than the one in the ad. On a typical hospital website, with hundreds of doctor profiles, package pages, branch pages and news items, the choice can be right, merely adequate, or plainly wrong for a patient who needs an appointment.

The article is a teardown method. You will audit a test campaign's landing URLs, classify the errors, and decide which are fixed with exclusions, which with page changes and which by switching the feature off. It does not report results from any named hospital. The decisions come from what Google documents and from the structure of hospital websites.

If your landing pages are weak before AI Max, expansion will find the weak ones faster. Our guide to healthcare landing page elements and the specialist referral page checklist cover page quality; here we cover which page gets chosen.

What final URL expansion does, per Google

Google's help page on AI Max for Search campaigns says final URL expansion sends traffic to the most relevant URLs on your site when that is likely to perform better. It uses only URLs relevant to the query and the ad group theme. It requires text customization to be on, and pinned assets are not respected when a different URL is chosen. If an ad group has a pinned headline for a regulatory statement, a different landing page may serve without it.

The controls the page lists are the ones you will use in an audit:

• URL exclusions at campaign level, which stop specific URLs serving as landing pages.
• URL inclusions at ad group level, which add URLs that expansion did not capture.
• Reporting: a "Selected by" column in the landing pages report, and a combined view of search terms, headlines and URLs.

The same page warns that dynamic landing pages can break with certain tracking templates, which cause 404 errors, and recommends using the lpurl tag in compatible patterns and testing landing pages before enabling AI Max.

Why hospital sites are a hard case

A hospital site differs from a retail catalogue in four ways that matter here.

1. Doctors move. A profile page may exist after a consultant has left, changed branches or stopped taking appointments. A query for that doctor's specialty could match the old page.
2. Packages expire. Health-check and surgical packages are priced and withdrawn. Old pages often stay live and indexed.
3. Branches overlap. The same service appears on five location pages. Expansion may choose the wrong city.
4. Content pages look like service pages. A long educational article on knee pain can match a procedure query better than the booking page does, but it has a weaker path to an appointment.

Each of these is a real patient problem, not only an advertising inefficiency. A person who clicks an ad and arrives at a page for a doctor who is no longer there will lose trust in the hospital.

Step 1: Export the landing pages

Run the audit on a test campaign, or on a campaign that has had final URL expansion on for at least one full conversion cycle. Pull the landing pages report with the "Selected by" column. Export the URL, clicks, cost and conversions. Add the original ad group final URL for comparison.

Then add two columns from the website and CRM side:

• Page status: live and current, outdated, duplicate, or not meant for paid traffic.
• Enquiry outcome: qualified enquiries and booked appointments attributed to that URL, from your lead log.

If your enquiry log does not record the landing page, add a hidden field that captures it on forms and in click-to-call tracking. Without that, the audit measures clicks, not patients.

Step 2: Classify each expanded URL

Create a simple table and put every URL selected by expansion into one of five classes.

Count how many URLs and how much spend fall in each class. The proportion is the output of your audit. Do not assume it will match another hospital's.

Step 3: Exclude, include or fix

Exclusions are the fast fix. Add every outdated, wrong-intent or non-paid-traffic URL as a campaign-level exclusion. Because URL exclusion works on specific URLs, a rule based on a directory pattern, where the interface supports it, saves effort on a site with many profile pages. Check the options in your account, because the available matching rules are set by the interface and I have not confirmed their detail from Google's documentation.

Inclusions work in the other direction. If the system never selects the best procedure page, add that URL at ad group level so it becomes eligible.

Fixing pages is the durable answer. Three fixes pay off repeatedly.

• Retire stale pages properly. Redirect withdrawn packages to the current equivalent and mark departed doctors' pages so they are not offered in paid traffic.
• Give each branch page a clear location signal. City or locality in the heading and in structured data helps people and systems tell the pages apart.
• Keep one canonical page per service per branch, and make sure the call and booking actions are present on it.

Page feeds and what they can and cannot do

Teams sometimes ask whether a page feed can steer expansion. Page feeds are a Dynamic Search Ads tool for listing the URLs you want eligible. Whether a feed constrains AI Max's final URL expansion in your account is something to check in the current interface and documentation; I could not confirm it from the AI Max help page, which lists URL exclusions and inclusions as the controls. Plan around exclusions and inclusions, and treat a feed as an optional extra only if your account supports it.

Doctor pages: a special rule

For individual doctors the stakes are higher because of advertising rules. In India, what a doctor's page may say and how it may be promoted falls under NMC professional conduct and advertising regulations, and our NMC compliance handbook explains the framework. In the UAE, health advertising is subject to regulator permits. Paid traffic to a doctor profile that carries claims your compliance owner would not approve in an ad adds a risk you did not intend to take.

A conservative approach is to exclude individual doctor profile pages from AI Max expansion by default and send doctor-intent queries to a department or specialty page that lists approved clinicians. Add doctor profiles back one at a time after review.

Test the experience, not just the report

An audit of URLs in a spreadsheet misses what the patient sees. For the ten highest-spend selected URLs, do the following on a mid-range phone:

1. Open the page from a search result on mobile data.
2. Time how long it takes to show the phone number and the booking button.
3. Confirm the page names the service searched, the branch and the next step.
4. Submit a test enquiry and confirm it reaches the team with the landing page recorded.
5. Check that tracking parameters survive a redirect, because broken parameters hide conversions.

If you use call tracking, check that the dynamic number swap works on every page that expansion can select, not only the page you designed. Our call tracking and attribution service page covers the setup.

When to switch final URL expansion off

Switch it off when the audit shows that most spend goes to wrong-intent or outdated pages and you cannot fix or exclude them at the pace the account needs. Switching off text customization also switches off final URL expansion, but note it also removes generated headlines. Weigh that trade; many hospitals prefer to keep search term matching and fix URLs by hand.

Leave it on when the selected pages are consistently as good as, or better than, the pages you chose, and the exclusion list is stable.

Where to start

1. Export the landing pages report with the "Selected by" column.
2. Add page status and enquiry outcome columns.
3. Classify each URL and total the spend per class.
4. Exclude outdated, wrong-intent and doctor profile URLs.
5. Add inclusions for the best procedure pages.
6. Redirect or retire stale pages on the website.
7. Repeat the export after the next full conversion cycle and compare classes.