Menu
← Directory data & sourcesBARE RESEARCH · OCTOBER 2026
A SMALL STUDY, WITH ITS LIMITS IN VIEW

What a hosting lookup can really tell you.

Finding a network is easy. Naming the company a website owner pays is harder. Our first pilot shows why those answers belong in separate boxes.

Observed 19 October · Published 20 October 2026
1,000seed domains checked
280confirmed active in the bounded check
97eligible public research websites
LATEST RULE UPDATE · 21 OCTOBER 2026

21 / 97 sites have a likely host

One more public service clue: Webflow’s documented CNAME identifies its platform behind Cloudflare. Replaying the same saved observations increases coverage from 20 to 21, with medium confidence and no lost or switched labels. Network and CDN observations stay separate.

What this update establishes

The directory still uses the original 19 October observation dates. A separate passive recheck also found the Webflow route and matching page fingerprints, but its fresh HTML is not combined with the older network data. CMS and copied builder assets alone do not establish hosting. These are coverage counts, not verified customer relationships or an accuracy rate.

Explore the hosting observations ↗
Earlier DNS and network review: 16 → 20 likely hostsEARLIER DNS / NETWORK UPDATE · 21 OCTOBER 2026

20 / 97 sites have a likely host

Up from 16, using the same saved observations. Reviewed DNS and platform clues paired with the serving network add 4 medium-confidence inferences. The network and CDN stay separate. No new crawl or paid API calls.

How the matching works

A documented provider DNS or platform association must agree with the same sampled serving network. DNS rules require all nameservers to agree; service rules require a specific reverse-DNS or response fingerprint. All require the exact network organization, ASN and domain within the original validity window. Mismatches and CDN routes stay unconfirmed. A reseller may still be involved.

These are coverage counts, not a proven accuracy rate. The original study and chart below preserve the first release. Explore the current hosting observations ↗

ORIGINAL RELEASE · 20 OCTOBER 2026

One website. Different layers.

Each bar counts sites with that observation. A site can appear in more than one bar.

Likely hosting company11 / 97

A specific platform or hosting service was inferred.

CDN observed47 / 97

An edge provider; this does not reveal the origin.

IP network identified97 / 97

The network operator, not necessarily the retail host.

Cloud infrastructure4 / 97

A matching infrastructure clue, kept separate from hosting.

The 11 hosting inferences have medium confidence. These are coverage counts, not independently verified accuracy or hosting market share.

Download the chart

What we checked

The fixed pilot contained 250 Majestic Million seed domains and 750 Common Crawl seed domains. A bounded DNS and homepage check on 19 October found 97 and 183 active sites respectively: 280 in total.

“Active” means enough accessible, substantive HTML was returned by this check. Robots exclusions, challenges, transport failures, sparse pages and inconclusive results are not proof that a website is inactive.

This public analysis uses only the 97 eligible Majestic-seed observations. The Common Crawl cohort remains outside the published directory while its redistribution basis is reviewed.

What the result means

A CDN can conceal the origin. A cloud network can serve hundreds of hosting brands. DNS can be managed by a different company. Joining those labels into one confident host name would hide the uncertainty.

Our public projection was recomputed without registry, BGP, RPKI or RIPEstat inputs. It reuses existing observations, with no fresh crawl and $0 in additional paid API or data costs. Existing compute and storage are not priced in that figure.

This is a small, selected sample. It cannot measure worldwide provider popularity, overall detection accuracy or changes over time.

Use the finding, keep the context.

Share this dated study, or reuse its chart with attribution. Credit Bare Audit / Todorov.co, IPinfo Lite and Majestic Million; link to this study and the licence, note changes and keep required ShareAlike terms.

Sources and CC BY-SA 4.0 reuse details ↗ · Current eligible research data (JSON) ↗

The chart preserves the original release counts. Current directory records expire 31 days after observation and may also be removed, so today's download can contain fewer websites. Later rule reviews may also change its inferred hosting labels.