Checking sign-in…
Providers
Which companies health facility domains depend on. A DNS operator runs a domain's name servers; a registrar holds its registration; one company can be both. Concentration is the plain picture for reporting.Analysis is for picking 2–3 providers to interview about whether a provider-level notification could work in a later phase. No provider is notified in this phase.
DNS operators
Loading…
Registrars
Loading…
| # | Company | Brands | Domains | Share | Cumulative |
|---|---|---|---|---|---|
| Loading… | |||||
How these numbers are counted
- Share = domains that depend on the company ÷ all domains counted (every scanned domain, or the Phase 3 domains).
- Double counting: a domain that uses name servers from two different companies counts under each, so DNS operator shares can add up to more than 100%. This is the only place a domain is counted twice.
- Not double counted:
- A domain using two brands of the same company (say, Network Solutions and Bluehost name servers) counts once for that company.
- Registrars: a domain has only one registrar, so registrar shares never overlap and add up to at most 100%.
- The total of domains counted (the figure each share is divided by) counts every domain once.
- Grouped by company: brands owned by one company are added together (listed in each row). Ownership comes from a list we keep by hand, shown below. A company or brand not on it is counted under its own name.
- Scanned domains: domains whose latest DNS scan is complete. Companies with fewer than 30 of them are counted under "All others".
- Dependency only: this view counts which companies domains rely on. It says nothing about how well each company's domains are set up; that's the Analysis view.
Company ownership list
Companies whose brands are grouped or renamed here. Ownership last checked —; companies buy and sell brands, so it's re-checked before each report.
| Company | DNS operator brands | Registrar brands |
|---|
How to read this. A big provider has more problem domains simply because it has more customers. So this view compares rates: the share of a provider's domains with a problem, next to the same share for everyone else. A rate well above everyone else's points at something about that provider (a default, a platform design, a shared setup) rather than many separate customer mistakes. It's a lead to check, not proof: the checks under each provider say what might explain it away.
Each dot is a provider. The dashed line is the rate across all scanned domains; the shaded funnel is how far a provider of that size could drift from it by chance. Dots above the funnel have a higher rate than chance explains; dots below it do better than everyone else. Only the biggest and the flagged providers are named; hover over a dot for its name, click it to open it in the table, and drag across the chart to zoom in.
Click a provider for a plain-English summary, the problems where it stands out, and the checks. Click a column heading to sort. Tick up to 3 to compare.
| Compare | Provider | Domains | Rate here | Everyone else | × everyone else | Stands out on | Phase 3 (risk 1–2) |
|---|---|---|---|---|---|---|---|
| Loading… | |||||||
Brands are kept apart here (Network Solutions and Bluehost are both Newfold Digital) because platforms behave differently; Concentration groups them by company.
How these numbers are counted
- Rates, not counts: a provider's rate = its domains with the measure ÷ all its domains. "Everyone else" is the same rate across every other provider combined.
- Every scanned domain is used to compute rates. The Phase 3 domains are too few to tell a real difference from chance; the Phase 3 column only shows what a change at that provider would reach.
- Range: the 95% range under each rate is where the true rate very likely lies, given how many domains the provider has. Small providers get wide ranges. Providers with fewer than 30 domains aren't rated.
- Higher / Lower: only when the provider's whole 95% range is above (or below) everyone else's rate and its rate is at least double (or at most half) theirs. A real but smaller gap shows as "Same range" and the summary says it's a bit higher or lower.
- Stands out on: a problem the provider has at least twice as often as everyone else, on 5 or more of its domains, with its 95% range above everyone else's rate.
- Double counting: a domain that uses name servers from two DNS operators counts under each.
- Problems = findings that our findings policy would put in a letter; a domain with several counts once in a rate.
- Name servers don't answer: a DNS operator whose name servers our scans find unanswering on at least half its domains (hollow dot on the chart). That usually means nobody is running DNS there (the name server domain lapsed, was abandoned or is parked), though it can also be an operator blocking our scans. It doesn't say anything was intentional. Either way its rate reflects broken delegations, not a provider practice, so it is no lead for an interview until someone has looked. This comes from our scans only; we don't check whether the name server domain is for sale.
- Together: if 80% or more of the domains with one stand-out problem also have a second problem, the checks say so. Two findings that nearly always come together are probably one underlying problem (for example, name servers that answer neither UDP nor TCP), so count them as one when talking to the provider.
- Name server setup: asks whether the provider's problem domains share one particular setup. A setup is the domain's list of name servers, with numbers written # (so connect1 and connect2.squarespacedns.com are one setup) and the domain's own name written <own> (so no facility name shows). The checks name a setup only when all three hold: at least 10 problem domains use it, it covers at least half the provider's problem domains, and that is at least twice its share of all the provider's domains. Example (November 2026): 99% of Squarespace's risk 1–2 domains use connect#.squarespacedns.com, which only 20% of its domains use, so ask Squarespace about that setup, not about everything. Fewer than 10 problem domains on any one setup: "too few to tell".
- These thresholds are a working draft and may change.