What does a CrUX popularity rank actually measure?
Part of US and Canada targeting
The short answer
A CrUX popularity rank is a coarse bucket, such as top 10k or top 50k, that Google assigns to a website origin from the total navigations Chrome users make to it. A rank says how busy an origin is relative to others in that dataset, not how many people visit it or will see an ad.
Key takeaways
- Google defines the rank as a relative measure of popularity within the CrUX dataset, based on total navigations to an origin.
- Each bucket excludes the one above it, so top 5k holds roughly 4,000 origins and top 10k roughly 5,000.
- Country ranks come from country datasets in BigQuery, with eligibility applied separately for each country.
- A bucket cannot be turned into visitors, impressions or province level reach, because Google publishes no conversion.
- BigQuery country data changes once a month; the CrUX API refreshes daily but offers no country dimension.
How does Google define the CrUX popularity rank?
Google's CrUX metrics page (opens in a new tab) lists popularity under its experimental metrics and describes it as "a relative measure of site popularity within the CrUX dataset, measured by the total number of navigations on the origin." Three parts of that sentence matter to a planner. Relative means the rank only places an origin against other origins. Navigations means page loads, not people. Origin means the whole host, such as https://www.example.com, never a single article or section.
Definition
- CrUX popularity rank
- A coarse bucket in the Chrome UX Report that places a website origin among the most visited origins in the dataset, based on total navigations by eligible Chrome users. In BigQuery it is stored in the field experimental.popularity.rank.
The same page sets the scale: "Rank is on a log10 scale with half steps (e.g. top 1k, top 5k, top 10k, top 50k, top 100k, top 500k, top 1M, etc.), with each rank excluding the previous (e.g. top 5k is actually 4k URLs, excluding top 1k)." It adds that "The upper limit is dynamic as the dataset grows," and that popularity "is provided as a guide for broad analysis." The field name experimental.popularity.rank, and the existence of country specific ranks, come from Google's rank magnitude article (opens in a new tab), which predates the half steps.
| Bucket label | Origins it holds | What a planner can write |
|---|---|---|
| top 1k | The 1,000 busiest origins | Among the 1,000 most navigated origins in this dataset and month |
| top 5k | About 4,000 origins, ranked 1,001 to 5,000 | Busy, but outside the top 1,000 |
| top 10k | About 5,000 origins, ranked 5,001 to 10,000 | Between the 5,000th and 10,000th most navigated origins |
| top 50k | About 40,000 origins | Somewhere in a wide band of 40,000 origins; position inside it is unknown |
| top 1M and beyond | Hundreds of thousands of origins | Present in the dataset, with no finer position available |
Because buckets widen at every step, two origins in top 50k can differ greatly in traffic, and an origin can sit just inside one bucket while another sits just outside it. To pull every origin in the top 10,000 from BigQuery, the rank magnitude article tells analysts to filter on experimental.popularity.rank of 10000 or less rather than equal to 10000, since the 10k bucket alone omits the top 1,000.
How are CrUX country ranks calculated?
Country ranks live in country datasets. Google's dimensions documentation (opens in a new tab) says country level datasets are provided in addition to the global dataset, "with the standard eligibility requirements applied at a country level," and that "A table is provided for each country." The BigQuery reference (opens in a new tab) names them country_CC, so Canada is country_ca and the United States is country_us, alongside a materialized country_summary table that carries the country code as a column.
Google does not publish a formula beyond the metric definition. What its pages do state is enough to describe the method in plain words: the country is inferred from each user's IP address, only eligible users and origins count, and the rank reflects total navigations to the origin. Read together, a Canadian rank places an origin among origins with enough eligible Chrome traffic from Canada, measured by navigations from Canada.
Why can a CrUX rank not estimate how many Canadians will see an ad?
The rank counts navigations, groups them into buckets, and covers only part of the browsing population. Google's methodology (opens in a new tab) limits contributing users to people who enable usage statistic reporting, sync their browser history and have no Sync passphrase, on desktop Chrome or Chrome on Android. Chrome on iOS, Android apps using WebView and other Chromium browsers do not contribute.
| Planning statement | Supported by the bucket? | Reason in Google's documentation |
|---|---|---|
| The origin is among the busier origins for Chrome users in Canada this month | Yes | Country datasets rank origins by navigations from that country |
| The site reaches a stated number of Canadians | No | Navigations are page loads, and the rank is relative, with no published conversion to users |
| Readers are concentrated in one province or city | No | The finest geography is country; there is no province or city dimension |
| A given article or section is popular | No | Rank is measured for the origin as a whole |
| The publisher can deliver a stated volume of impressions | No | CrUX describes Chrome visits, not ad inventory or seller capacity |
| Popularity rose between two months | Only if the bucket changed | Movement inside a bucket is invisible |
Google also states that "A small amount of randomness is applied to the dataset to prevent reverse-engineering of sensitive data," while aggregate statistics stay accurate. For a planner the practical rule is simple: use the bucket as evidence that an audience exists in the country, and take any number of people, impressions or reach from a source that measures it directly. The hub guide explains where country evidence sits in a full Canadian or US plan.
Why do some sites have no CrUX rank at all?
An origin gets a bucket only if it is in the dataset. The methodology requires origins to be publicly discoverable and sufficiently popular, and says "An exact number is not disclosed" for the threshold. It also lists exclusions that catch real publishers: pages that fail discoverability, such as those returning a non 200 status after redirects or carrying a noindex directive, and origins or pages with more than 20 percent of their traffic excluded because of ineligible dimension combinations.
Before reading a missing rank as low popularity
- The origin queried matches the host readers actually land on, including www or no www, after any redirect.
- The origin was entered in lowercase with its scheme, as CrUX stores it.
- The query used the country table for the brief's market, not the global table, or the reverse.
- The month queried has been released, since each month appears only after the second Tuesday of the next.
- The site's main pages return a normal status and are not marked noindex.
- The absence is recorded as missing data in the plan, never as a rank of zero.
A country table can also lack a site that appears globally, because eligibility is applied per country. A small or local publisher may fall below a country threshold; the guide on small websites with no CrUX data covers what other evidence can support them on a plan.
How often is CrUX data updated, and which release has ranks?
CrUX has three public outlets on different clocks, and only one of them carries country data. The BigQuery dataset adds a new monthly table, and Google says it "is released on the second Tuesday of the following month," with yyyymm defined as the "Month of the data collection period." The CrUX API (opens in a new tab) serves "a 28-day rolling average of aggregated metrics," updated "daily around 04:00 UTC." The History API (opens in a new tab) "is updated each Monday around 04:00 UTC" and holds up to 40 weekly collection periods.
| Outlet | Refresh | Country data | Use in planning |
|---|---|---|---|
| BigQuery monthly tables | Monthly, second Tuesday of the next month | Yes, country_CC tables and country_summary | The source of country buckets; record the yyyymm month |
| CrUX API | Daily, 28 day rolling window | No, per the dimensions page | Confirms an origin is in the global dataset |
| CrUX History API | Weekly on Monday, up to 40 periods | Not offered; aggregated by form factor | Performance trends, not popularity by country |
Google says the API and History API have no service level agreement for update times. A rank recorded for a plan should therefore carry the month of the table it came from, not the day someone opened the file, so that a later reviewer can reproduce it.
What does reading a rank look like on a real brief?
Fictional example
A loose leaf tea retailer in British Columbia
An unnamed loose leaf tea and teaware retailer based in the Okanagan asks its agency for display placements beside tea brewing guides, food writing and home cooking articles, for adults in British Columbia. The planner shortlists twelve publishers by topic first, then asks an analyst for the latest released month of the country_ca table.
Three US food sites come back in the Canadian top 50k, so the planner notes them as US-based, popular in Canada. A Vancouver food magazine sits in a lower Canadian bucket than one of those US sites, and the account lead asks whether the US site therefore reaches more British Columbians. The planner answers that the data cannot say: buckets are national, relative and counted in navigations, and the magazine's provincial focus is the stronger fit for this brief.
Two small tea blogs have no Canadian record. The planner checks that their origins were entered correctly, marks them as missing data, and keeps them on the list on the strength of their articles. Province targeting is left to the buying platform's location settings.
How does PlacementGraph use country popularity ranks?
Frequently asked questions
Is the CrUX popularity rank the same as a site's position in search results?
No. Google defines the rank from total navigations by eligible Chrome users, whatever led to those visits. It is a dataset metric published as a guide for broad analysis, and Google's CrUX documentation describes it only in those terms, with no connection drawn to how a site appears in Google Search.
Do visits inside apps count toward a CrUX rank?
Only some. Google's methodology includes Android Chrome, including Custom Tabs and WebAPKs, but says Android apps using WebView do not contribute, nor does Chrome on iOS. A publisher whose readers arrive mostly through in-app browsers on social apps can therefore look less popular in CrUX than its own analytics suggest.
Can two sites in the same bucket be treated as equal on a plan?
Only as equally evidenced, not as equal in traffic. A wide bucket such as top 50k holds about 40,000 origins, so two sites inside it can differ many times over in navigations. Choose between them on content fit, language, seller path and the publisher's own aggregated figures.
Why did a site's Canadian bucket change from one month to the next?
The rank is relative, so a bucket can change because the site gained or lost navigations, because other origins moved, or because the upper limit of the dataset grew, which Google says is dynamic. Record both months and the change literally, and avoid describing it as a traffic trend.
Sources
- CrUX metrics: popularity (opens in a new tab)Chrome for Developers
- CrUX methodology (opens in a new tab)Chrome for Developers
- CrUX dimensions (opens in a new tab)Chrome for Developers
- CrUX on BigQuery (opens in a new tab)Chrome for Developers
- CrUX API (opens in a new tab)Chrome for Developers
- CrUX History API (opens in a new tab)Chrome for Developers
- Using the CrUX rank magnitude (opens in a new tab)Chrome for Developers
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.