Use readable URLs with words instead of opaque IDs or tracking parameters
Canonical URLs should look like /blog/ai-readiness, not /p?id=8c4e91a2-utm_source=x.
What this signal tests
We inspect the canonical URLs of pages on your site and look at each path segment. Segments that are word-based and hyphen-separated pass. Segments that are UUIDs, long base64 strings, twelve or more hex characters, six or more digits, or underscore-separated tokens are flagged. We also flag canonical URLs that carry tracking parameters like utm_source, fbclid, or gclid in their query string.
Why it matters for your visibility in AI
AI ranking and citation systems treat the URL itself as a strong relevance signal. Readable, descriptive URLs telegraph the page's topic before the assistant even reads the content; opaque ID-based URLs are pure noise. When two pages cover similar material, the one with /guides/api-rate-limits is dramatically more likely to be cited than the one with /p/8c4e91a2-1b3f-4e89. The concrete consequence is broken citations. Users sharing AI answers paste the URL into chats, social posts, and tickets. A readable URL survives that journey intact and self-describes its target; an ID-based URL becomes a random string that recipients have no reason to trust or click. Tracking parameters compound the problem by making the URL look spammy.
Pass criteria at a glance
| Criterion | Passes when |
|---|---|
| >=90% of canonical URLs pass readability checks. |
How we test it
We collect canonical URLs from sitemap entries and rendered link tags, then inspect each path segment. Segments are flagged when they match UUID patterns, base64-like strings of twenty or more characters, hex strings of twelve or more characters, numeric runs of six or more digits, or use underscores as word separators. We then check each URL's query string for known tracking parameters. Ninety percent or more of canonical URLs must pass.
Show technical detection method
For canonical URL paths flag segments matching: UUID, base64-ish >=20 char, >=12 hex, >=6 digit numeric, underscore separators. Flag utm_*, fbclid, gclid in canonical query.
If your site fails: how to fix it
- Configure your CMS to generate slug-style URLs from the page title; in WordPress, set Permalinks to "Post name" rather than the numeric default.
- For existing pages with opaque IDs in the URL, set up 301 redirects from the old URL to a new readable slug, and update internal links to point to the new path.
- Strip tracking parameters from canonical link rel="canonical" values; tracking belongs on inbound campaign links, not on the canonical URL the site declares for itself.
- Use hyphens, not underscores, to separate words in slugs (api-rate-limits, not api_rate_limits); hyphens are the convention crawlers expect.
- For e-commerce SKUs that must appear in URLs, combine the SKU with a descriptive slug, for example /products/wireless-headphones-XQ-300 rather than /products/sku/8847291.
Quick facts
| Maturity | ESTABLISHED |
|---|---|
| Weight | medium |
| Category | Content Clarity |
Primary sources
Related signals
Frequently asked questions
Do tracking parameters hurt my SEO too?
On a canonical URL, yes. Search engines may treat each tracking variant as a separate page and dilute ranking signals across them. A clean canonical URL with tracking confined to inbound links (which set a 301 to the canonical) keeps both classic SEO and AI ingestion clean.
What if my legacy URLs are full of IDs and I cannot change them?
Adopt readable slugs for all new content immediately, and progressively migrate the most-linked legacy pages with 301 redirects. You will not fix everything at once, but the trend matters - new content shipping with clean URLs gradually shifts your overall score upward.
Is there a maximum URL length?
There is no hard cap, but extremely long URLs (over about seventy-five characters) are harder to read, share, and remember. Aim for short, descriptive slugs of two to five words per segment. Length is not the failure mode the signal targets, but it correlates with readability.
Run your own scan
Run a free scan and see how your site grades across all 155 AI-readiness signals.