Guide — traffic proxies

Traffic proxies for publishers: CrUX presence vs third-party estimates

When you screen hundreds of thousands of publisher domains, you need a traffic signal you can reproduce and defend — not a single invented visitor count. This guide compares two reproducible proxy approaches: Google's Chrome UX Report presence and rank groups, and third-party traffic-estimate vendors. Both are legitimate data sources; the question is which proxy fits screening at scale.

Hundreds of thousands
domains a proxy must scale to per vertical (estimate)
2
proxy families compared: CrUX vs estimate vendors
0
invented visitor numbers in our deliverable
First, the framing

Why a proxy at all?

No one outside a publisher's analytics has their true visitor count. Everyone else works from a proxy — a measurable signal that correlates with traffic. The honest move is to name your proxy, report it as a proxy, and make sure anyone could reproduce it. That is what keeps a screening deliverable auditable.

Reproducible

A good proxy can be re-derived by a third party from a stated source. If a number can't be reproduced, it can't be checked — and we won't ship it as fact.

Comparable

At screening scale you need the same signal on every domain, so publishers rank against each other consistently — not a mix of sources per row.

Honest about uncertainty

A proxy is labeled as a proxy. Presence-and-rank-group framing communicates a band of confidence rather than a false-precise visitor figure.

The two data sources

CrUX presence and rank groups vs traffic-estimate vendors

Both are real, widely-used data sources with legitimate uses. They answer slightly different questions, and each has a place in a screening stack. Here is what each actually is.

Google CrUX presence + rank groups
The Chrome UX Report publishes real-user experience data for origins that meet a popularity threshold in Chrome.
Presence itself is a signal: an origin appearing in CrUX has cleared a minimum audience bar.
Google reports popularity as coarse rank groups (broad magnitude bands), not exact visitor counts.
Directly reproducible from a public Google dataset — anyone can re-check a domain.
Chrome-user based, so it reflects that population; very small sites may not appear at all.
Third-party traffic-estimate vendors
Vendors such as commercial traffic-analytics providers model visit estimates from panels, clickstream, and other inputs.
They return a specific estimated visits figure, which is convenient for a single headline number.
Useful for competitive research and deeper single-site analysis where their tooling shines.
Methodologies are proprietary, so a given number is not independently reproducible from a public source.
Coverage and licensing at whole-vertical scale differ from a public dataset lookup.
Not a verdict on quality: traffic-estimate vendors are strong tools for their intended jobs. The point here is narrower — which proxy is reproducible and comparable across a whole-vertical screen of hundreds of thousands of domains.
CrUX presence

Whether a domain's origin appears in the public Chrome UX Report — a binary signal that the site has cleared Chrome's minimum-popularity threshold. Absence is informative too, especially for filtering the very smallest sites.

Rank group

The coarse popularity band Google assigns an origin (a broad magnitude bucket rather than an exact rank). It orders publishers by scale without claiming a precise visitor count.

Side by side

Comparing the proxy approaches

The same domains, viewed through each proxy, for the specific job of screening a whole vertical. Neither column is "wrong" — they optimize for different things.

 CrUX presence + rank groupsTraffic-estimate vendors
What it returnsPresence flag + coarse rank group (magnitude band)A specific estimated visits figure
Reproducible from a public sourceYes — public Google datasetNo — proprietary modelled estimate
Comparable across all domainsSame signal on every origin presentDepends on coverage per site
Precision claimedDeliberately coarse (a band)Precise-looking single number
Best-fit useWhole-vertical screening & rankingDeep single-site & competitive analysis
Coverage of tiny sitesMay be absent below the thresholdVaries; often modelled or blank
Our discipline

Why we report reproducible proxies, not invented numbers

Across a run that can screen hundreds of thousands of domains, the traffic field has to be defensible on every row. We default to CrUX presence and rank group because any client can re-derive them from a public source — and we never ship a visitor count we can't stand behind.

0
domains screened in one reference run
Public
source behind the traffic proxy
Band
not a false-precise single figure
0
invented visitor numbers shipped

If your workflow needs vendor estimates as an added field, we can incorporate a licensed source alongside the reproducible proxy — clearly labeled as a modelled estimate, never presented as measured fact.

Practical guidance

Which proxy for which job

The two data sources are complements more than rivals. Match the proxy to the decision you're making.

Screening & ranking a whole vertical

Use CrUX presence and rank group. You get one reproducible, comparable signal on every domain, so publishers sort by scale consistently and the field survives an audit.

Deep-diving a shortlist

Bring in a traffic-estimate vendor once you're evaluating a handful of finalists. Their richer single-site tooling adds context where an exact-looking figure and trend view are worth the proprietary trade-off.

The rule we hold: whichever proxy is on a row, it is labeled for what it is — a reproducible presence-and-rank signal, or a modelled vendor estimate. The reader always knows which they're looking at.
Questions

Traffic proxies — FAQ

Is CrUX a traffic number?
Not exactly. CrUX is Google's public real-user experience dataset for origins that clear a popularity threshold in Chrome. We use two things from it as a traffic proxy: whether a domain is present at all, and its coarse rank group. Both are reproducible from a public source; neither is a precise visitor count.
Are traffic-estimate vendors inaccurate?
They are legitimate, widely-used tools, strong for competitive research and deep single-site analysis. Our preference for CrUX at screening scale isn't a judgment on their quality — it's about reproducibility: a modelled estimate can't be independently re-derived from a public source, which matters when a field appears on every one of hundreds of thousands of rows.
Why not just give me one visitor number per site?
Because we don't have direct access to any publisher's true analytics, and neither does anyone else outside the site. A precise-looking number implies a certainty we can't back. A presence flag and rank-group band communicate scale honestly and can be checked by anyone.
Can you include vendor estimates if I want them?
Yes. If your workflow relies on a specific traffic-estimate vendor, we can add a licensed source as an extra field alongside the reproducible proxy. It's clearly labeled as a modelled estimate so it's never confused with the auditable CrUX signal.
What traffic fields are in the deliverable?
Each row carries CrUX presence and rank group as the default traffic proxy, next to the ICP verdict, per-signal confidence, language, and geo focus. It imports straight into a partner CRM, and every traffic value is labeled for exactly what it is.

Screen your vertical on signals you can defend

The free pilot runs the full pipeline on your category and delivers your first 20 qualified publishers — each with a reproducible traffic proxy — plus a full-run projection. No cost, no obligation.

Request a Free Pilot