Should citation reports count domains and pages separately?

Definition: Yes. A citation report should distinguish unique source URLs, which identify linked resources, from unique domains, which group those URLs by domain name. Repeated links to the same URL form a separate measure of citation occurrences.

Key facts

• The Source Coverage and Citation Bias study counts each URL once within a response and analyzes unique URLs and domains separately.
• RFC 3986 treats URL schemes and hosts as case-insensitive, while other generic components are assumed case-sensitive unless specified otherwise.
• The Public Suffix List explanation identifies registration boundaries such as co.uk, which matter when grouping subdomains.

What does the study count?

In its December 2025 version, Source Coverage and Citation Bias examines user-facing citations. A URL repeated within an answer contributes once to that answer's unique URL count. The study reports means of 4.3 URLs and 3.4 domains per AI search response in its sample.

These units answer different questions. Unique URLs measure the breadth of linked resources; unique domains measure breadth after grouping links by domain. Several pages from the same domain increase the URL total while contributing a single domain.

For reporting, retain citation occurrences alongside these deduplicated counts. State the aggregation boundary: a unique URL per response, a unique URL across the collected responses, and the number of responses containing that URL are different measures. Deduplication within an answer preserves the ability to measure recurrence across answers.

When do different URLs count as the same page?

A page count needs an explicit normalization rule: a procedure for mapping equivalent URL forms to a common identifier. RFC 3986's normalization rules give this equivalent pair:

HTTP://www.EXAMPLE.com/
http://www.example.com/

Lowercase the scheme and host when constructing a normalized identifier, and preserve the original link for inspection. Avoid lowercasing an entire URL: the RFC assumes other generic components are case-sensitive unless specified otherwise. Its query component also helps identify a resource, so removing every query string can merge distinct resources.

Record a fragment, the component introduced by #, separately if grouping links to sections of the same document. The RFC describes fragments as identifying secondary resources, including portions of a primary resource.

As of October 2026, Google's canonical URL documentation describes rel="canonical" as a strong signal for selecting a preferred URL among duplicate or very similar pages. A report that merges links using canonical annotations should label the resulting measure as canonicalized pages and retain the observed URLs and mapping. This makes the merge reviewable.

How should domains be grouped?

Specify whether a domain count uses the full hostname or the registrable domain, the name registered beneath a public suffix. Grouping by hostname preserves subdomain distinctions; grouping by registrable domain combines subdomains under that registered name.

The Public Suffix List explains why suffix boundaries require a list: registration policies vary, including registrations beneath co.uk. Taking only the final labels of every hostname can therefore produce incorrect groupings. A reproducible report records its grouping rule and the list version used.

Related terms

• Citation tracking
• Canonical URL
• URL parameters

FAQ

Should I count a repeated link more than once?

For a unique URL count per response, count it once, following the study's methodology. You can retain a separate occurrence count to describe repeated citation placement.

Can I remove all URL parameters before counting pages?

Keep them unless your normalization rule establishes equivalence. RFC 3986 specifies that query data participates in identifying a resource, so blanket removal can combine different pages.

Sources

1. arXiv, "Source Coverage and Citation Bias in LLM-based vs. Traditional Search Engines", 10 December 2025
2. RFC Editor, "RFC 3986: Uniform Resource Identifier (URI): Generic Syntax", January 2005
3. Public Suffix List, "Learn more about the Public Suffix List", undated
4. Google Search Central, "How to specify a canonical URL with rel="canonical" and other methods", updated 10 July 2026

---

Check how a page reads to answer engines. Open the GEO score analyzer