Core Web Vitals in Russia in 2026: country-level CrUX data and its limits

July 2026 data, CrUX field measurements via the HTTP Archive report. The figure is the share of origins where all three metrics fell within the threshold:

A nine-point lead on desktop sits well above the world level, and the explanation is likely simple: Russian wired and home connections are on average faster than in the countries dragging the global median down.

By individual metric, Russian mobile against desktop: LCP 65.56% and 82.21%, CLS 83.80% and 83.16%, INP 76.49% and 98.10%, TTFB 44.73% and 65.48%. The narrowest point on both devices is TTFB — server response speed, not layout.

Mobile trend, July 2025 → July 2026:

Russia improved — but three times more slowly than the world. A year ago it was ahead by 0.53 points; now it is behind by 1.58. The gap opened not through decline but because everyone else grew faster.

Among the 62 countries with at least fifty thousand eligible origins, Russia ranks 34th. Its neighbours on the list: Japan 71.87%, Germany 68.35%, Poland 62.90%, Ukraine 59.90%, Turkey 56.24%, Russia 51.63%, Kazakhstan 42.36%, Brazil 35.24%, India 31.70%.

By median value the gap looks smaller than by pass rate: the median Russian origin has a mobile LCP of 2,100 ms against 2,000 ms globally, and an INP of 125 ms against 100 ms. On desktop Russia is actually better on LCP — 1,500 ms against 1,600 ms. Caveat: these come at bucket resolution (100 ms for LCP, 25 ms for INP), so they are not exact values but the bucket the median origin falls into.

CrUX collects only from Chrome, and its methodology explicitly excludes Chrome on iOS, WebView applications and other Chromium browsers. Yandex Browser is built on Chromium but is not Chrome — its users do not enter these figures at all.

For Russia this matters more than for any other country on the list: Yandex Browser’s share there is nothing like its global share. How far this skews the result cannot be said — there is nothing to compare against, because no second source of field data for Russia exists.

A second caveat concerns the word "Russia" itself. The country is determined from the user’s IP address, not from the site. So this is "the experience of people located in Russia, on whatever sites they visited", not "Russian websites". A German site with enough Russian traffic is in this population; a Russian site without enough Chrome traffic is not.

Third: the Russian sample is shrinking fast. Eligible mobile origins came to 363,352 against 408,798 a year earlier — down 11.1%; desktop down 18.7%. The global sample lost 1.5% over the same year. Part of the year-on-year movement may reflect a change in sample composition rather than a real change in speed, and public data cannot separate the two.

And fourth, a technical one: INP is not measured everywhere. Of 363,352 Russian origins, 255,799 have INP data — 70.4%. An origin without INP data passes on two metrics instead of three. So roughly a third of Russian "passing" origins were never tested on INP at all.

Do not confuse them with lab scores. HTTP Archive also publishes median Lighthouse scores: for Russia, 0.455 on mobile and 0.615 on desktop. Reading those as "the performance of the Russian mobile web" is wrong: the crawl runs from datacentres in the United States on emulated Android. It is a synthetic measurement of the same sites from across an ocean, not anyone’s experience in Russia. Only the figures in the first two sections describe real users.

Do not add them to Google’s own summary. The CrUX release notes for July 2026 give 55.7% across 18,059,068 origins — a single count with no device split. HTTP Archive counts mobile and desktop separately on its own subsample. These are different quantities and must not be averaged. Google publishes nothing by country at all — every Russian figure here comes from the CrUX BigQuery extract via HTTP Archive, not from Google directly.

FID is gone. It was replaced by INP in March 2024, and in 2026 data every FID row is empty. Any source quoting FID for 2026 is working with dead data.

There is no 2026 Web Almanac. The latest edition is 2025, and its data is the July 2025 crawl. Its performance chapter carries no country breakdown whatever; Russia is not mentioned.

Other limits: origins are counted, not pages or traffic — a site with ten visits weighs as much as one with ten million; tablets are excluded; only home pages are tested; and a small random adjustment is applied so the data cannot be reversed.

First and foremost: if your audience is Russian, benchmarking against the global median is wrong. On desktop you compete against 69%, not 60%; on mobile against 52%, not 53%. Google’s "good" threshold is the same either way, but your competitors are not.

Second, the weakest metric in Russia on both devices is TTFB (44.73% on mobile). That is not about layout or images but about how fast the server answers. What works here is server-side caching, proximity to the user, and having a CDN — itself a story in the Russian web.

Third, the median page in the Russian sample weighs 2.33 MB on mobile, of which 750 KB is images and 662 KB scripts. That is a lab measurement, but weight is one quantity it measures reliably, and it feeds directly into LCP.

To check your own values, use the speed test; to see what is slowing the server response, the header check.

Sources: HTTP Archive technology report, CrUX release notes, CrUX methodology.