PageSpeed Insights "not enough data" / falls back to origin-level data

What’s actually happening

You run a specific URL through PageSpeed Insights expecting Core Web Vitals for that page, and instead the Field Data panel either shows nothing or quietly switches to "all user experiences on this origin." The origin numbers cover every page on the domain mixed together, so a fast landing page can look bad because the rest of the site drags the 75th percentile down. New pages and anything below a few thousand visits a month hit this constantly. The Lab Data section still works fine, which trips people up — Lighthouse always runs, field data does not.

Common causes

• The exact URL doesn't have enough real Chrome visits (roughly ~1,000 samples per device type per month) to clear the CrUX reporting threshold, so PSI drops to origin-level data
• The page is brand new or recently changed its URL, so CrUX hasn't accumulated 28 days of qualifying samples yet
• Most of the page's traffic comes from browsers or sessions CrUX doesn't collect from (non-Chrome, opted-out, enterprise policy, incognito)
• The page is gated, paywalled, or noindex and never reaches enough public Chrome users to qualify
• Mobile and desktop are counted separately — a page can have desktop field data but no mobile, or vice versa

How to fix it

1. Confirm it's a data-volume issue, not a setup problemRead the exact PSI message. "Falling back to aggregate data for all user experiences on this origin" means the origin qualifies but the URL doesn't. No field section at all means even the origin is below threshold. Neither is a bug you can fix in code — it's a traffic problem.
2. Lean on Lab Data while field data is missingThe Lighthouse run at the bottom of the report gives you LCP, CLS, TBT, and Speed Index for that exact URL right now. Use it to find and fix problems. Field data will follow once traffic accrues; lab data is your feedback loop in the meantime.
3. Check the URL in the CrUX API or BigQuery directlyQuery the CrUX History API for the URL. A flat "record not found" confirms it's below the page-level threshold. The public CrUX BigQuery dataset is origin-only, so don't expect URL rows there — use the API for per-URL checks.
4. Group similar pages and judge them togetherIf individual product or article URLs will never get enough traffic on their own, evaluate them as a template using lab tests on representative URLs, and watch the origin-level CrUX trend to catch sitewide regressions.
5. Drive more real traffic, then wait out the windowThere's no override for the threshold. Once the page gets enough Chrome visits, URL-level data appears on a 28-day rolling basis — so budget about a month after a traffic increase before expecting page-level numbers.