Configuring Excluded URLs

AI Search pulls answers from your school's web content. That's powerful — but only when the right pages are in the mix. Staff directories, faculty bios, graduate program pages, and internal news archives can all muddy a student's answer if they're left in scope.

Wildcards in Excluded URL paths let you draw a clean boundary around what AI Search can and can't pull from. This article shows you how to use them.

Where to find Excluded URL paths

Excluded URL paths live in your Halda Knowledge Base settings, just below the list of domains you've added. Anything that matches an excluded path is skipped — even if it lives on a domain that's otherwise in scope.

What the asterisk does

The asterisk (*) is a wildcard. It stands in for any characters, in any combination, anywhere in a URL. You can put it on either side of your pattern (or both) to control how flexible the match is.

Four patterns cover almost every case you'll run into.

1. Match anywhere in a URL: *keyword*

Put an asterisk on both sides of a word, and AI Search will exclude any URL that contains that word — no matter where it shows up in the path.

Example: *staff* excludes any URL containing "staff."

That catches yourschool.edu/staff-directory, yourschool.edu/about/staff, news.yourschool.edu/staff-announcements, and anything else with "staff" in the URL. Use this when you want to keep internal-facing pages out of student answers across your whole site.

2. Match a path that starts with something: /path*

Put the asterisk on the right side, and you're telling AI Search to exclude anything that starts with that path.

Example: /graduate* excludes any path starting with /graduate.

That covers /graduate, /graduate/admissions, /graduate-programs, and every page nested underneath. This is the right move for sections of your site that aren't relevant to your undergraduate audience.

3. Match across subdomains: *domain/section/*

Put asterisks on both ends with a path segment in the middle, and you can exclude a specific section across every subdomain you've added.

Example: *yourschool.edu/news/* excludes all news pages on any subdomain.

So www.yourschool.edu/news/..., admissions.yourschool.edu/news/..., and alumni.yourschool.edu/news/... are all skipped. Useful when news archives go back years and most of it isn't relevant to a prospective student asking about programs or application steps.

4. Exclude all subdomains: *.domain/*

Put asterisks on both ends of your domain with a . in front of the domain and a / following the domain. 

Example: *.yourschool.edu/*

This will ensure any subdomain associated with your main domain cannot be used when searching for content or serving information to a prospective student. 

What about entries without wildcards?

If you've already added paths that don't have wildcards, they still work. They behave as prefix matches — same as adding an asterisk on the right side.

You'll see a small convert button next to each one. Click it to switch the entry into explicit wildcard format. Worth doing if you want every entry to read the same way, but not required.

A practical workflow

When you're setting up exclusions for the first time, try this order:

1. Start broad with *keyword* patterns. Block anything you know shouldn't reach students — staff, faculty, internal, intranet, employee.
2. Add section-level exclusions with /path*. Graduate programs, alumni pages, donor sections, board minutes.
3. Layer in cross-subdomain patterns with *domain/section/*. Use these for news archives, events calendars, or anything that lives in multiple places.

After you save, ask AI Search a few questions a student might ask. If a stale or off-topic page still comes back, you know what to add next.

A few things to keep in mind

• Wildcards are case-sensitive in some configurations. If /Graduate* and /graduate* both exist on your site, add both to be safe.
• Be specific. *news* would block any URL with the word "news" — including, say, an admissions page about "good news for transfer students." *news/* or */news/* is usually safer.
• Exclusions stack. You can add as many patterns as you need. AI Search checks each URL against the full list.