Skip to main content
USA-Based Digital Agency

Original data · Free to cite

Webvello Research

Webvello Research measures how websites present themselves to search engines and AI assistants, and how well they are secured. We fetch the pages ourselves, publish the method and the raw counts, and state what each study cannot tell you. Every figure below links to its study, and the data is free to cite.

5 studies855 websites measured2 open datasets138 verified statistics15 rules tracked

New · November 17, 2026

30%

of the 100 largest U.S. banks’ domains have a registry lock; 88% rely on one DNS operator

How Well Do the 100 Largest U.S. Banks Protect Their Domains? A DNS Security Study

Read the study

Figures you can cite

Each sentence stands on its own, with its sample and date, so it can be quoted as written.

Headline figures from Webvello Research studies, with sample, date and source study
FigureFinding
29.8%28 of 94 domains of the 100 largest U.S. banks (29.8%) had a registry lock on November 17, 2026, falling to 5 of 45 among banks ranked 51 to 100.Domain and DNS security at the 100 largest banks
88.3%83 of 94 domains of the 100 largest U.S. banks (88.3%) relied on a single DNS operator on November 17, 2026.Domain and DNS security at the 100 largest banks
25.5%24 of 94 domains of the 100 largest U.S. banks (25.5%) were signed with DNSSEC on November 17, 2026.Domain and DNS security at the 100 largest banks
11 of 9411 of 94 domains of the 100 largest U.S. banks published no enforcing DMARC policy on November 17, 2026: 10 at p=none and 1 with no record.Domain and DNS security at the 100 largest banks
26.4%136 of 516 top-1,000 websites with a robots.txt (26.4%) disallowed at least one AI training crawler on November 13, 2026.AI crawler rules on the top 1,000 sites
14.5%75 of 516 top-1,000 websites with a robots.txt (14.5%) disallowed at least one AI search crawler on November 13, 2026.AI crawler rules on the top 1,000 sites
18.8% vs 9.5%Top-1,000 websites blocked OpenAI’s training crawler GPTBot about twice as often as its search crawler OAI-SearchBot: 18.8% against 9.5% (November 2026).AI crawler rules on the top 1,000 sites
22.9%134 of 585 top-1,000 websites (22.9%) published an llms.txt file on November 13, 2026.AI crawler rules on the top 1,000 sites
42 of 9742 of the 97 top-1,000 websites that disallowed GPTBot still allowed ChatGPT-User, the agent ChatGPT uses to open a page a person asks about (November 2026).AI crawler rules on the top 1,000 sites
66.3%63 of 95 California county and city homepages (66.3%) had at least one automatically detected WCAG 2.1 A/AA failure on October 17, 2026.Accessibility of California government sites
0 of 80None of 80 reachable marketing agency homepages disallowed an AI search agent in robots.txt (October 16, 2026).AI agent access on agency websites
3 of 80Three of 80 agency homepages refused at least one AI search or user agent while serving a browser normally, a sign of CDN or firewall rules (October 16, 2026).AI agent access on agency websites
40%32 of 80 agency homepages (40%) carried an Organization or LocalBusiness JSON-LD node with a name and contact details (October 16, 2026).AI agent access on agency websites
14%10 of 72 reachable marketing agency homepages (14%) shipped no JSON-LD structured data at all (July 30, 2026).Structured data on agency websites

Rules we track

Laws, standards and platform policies that shape websites, ads and AI tools, each with its primary source. Last checked November 16, 2026. Not legal advice.

  • AI calls and texting

    TCPA and AI voices (FCC 24-17)

    The FCC’s February 2024 Declaratory Ruling says AI-generated voices are “artificial” voices under the TCPA. Answering inbound calls is fine, but outbound AI calls need the person’s prior express consent.

  • AI calls and texting

    A2P 10DLC registration

    Texts sent from a business’s 10-digit number must be registered as a brand and campaign through The Campaign Registry. Carriers block unregistered traffic.

  • AI calls and texting

    Call recording consent (Cal. Penal Code §632)

    California requires every party’s consent to record a confidential call, so an AI receptionist’s greeting should tell callers the call is recorded.

  • Privacy and tracking

    California SB 690 (CIPA §638.51)

    Signed September 30, 2026. It removes the private right of action under CIPA §638.51, the pen-register and trap-and-trace section, for conduct on a website, online application or mobile app. It does not change §§631, 632 or 632.7.

  • Privacy and tracking

    CIPA wiretap and eavesdropping claims (§§631, 632, 632.7)

    Private lawsuits under these sections, with statutory damages of $5,000 per violation under §637.2, were left in place by SB 690. Courts have held that consent under §631 must come before the tracking, not after.

  • Accessibility

    ADA Title II web rule

    State and local governments must make web content and mobile apps meet WCAG 2.1 Level AA. An interim final rule published April 20, 2026 moved the deadlines; the standard and exceptions did not change.

See all 15 rules

Open datasets

Per-site results, free to reuse under CC BY 4.0 with credit to Webvello Research.

  • Domain and DNS security at the 100 largest banks

    96 rows · FDIC rank, bank, domain, DNS operators, DNSSEC, CAA, registry lock, DMARC, SPF, MTA-STS, certificate issuer, HSTS · November 17, 2026

    Download CSV
  • AI crawler rules on the top 1,000 sites

    585 rows · rank, domain, robots.txt and llms.txt status, verdict for 14 agents · November 13, 2026

    Download CSV

How we research

Five rules every study follows before it is published.

  • We collect the data ourselves

    Every study fetches live pages or files on a stated date. We do not repackage other people’s surveys as our findings.

  • The method is on the page

    Sample, tools, rules and dates are written out in enough detail for anyone to repeat the study.

  • Raw counts beside every percentage

    “26.4%” always comes with “136 of 516”, so a small sample cannot hide behind a big number.

  • Limits stated, not buried

    Each study lists what it cannot measure. A study about robots.txt does not claim to measure citations.

  • Data you can download

    Where the results are public facts about public sites, we publish the per-site CSV under CC BY 4.0.

Studies are led by George Shvaya, founder of Webvello. Our wider approach to measurement is on the methodology page.

How to cite Webvello Research

Quote the figure as written, name Webvello Research, and link to the study page so readers can check the method. Each study page has its own ready-made citation. Questions about a study or its data go through the contact page.

Credit line

Source: Webvello Research (https://www.webvello.com/research)

APA, latest study

Webvello Research. (2026, November 13). How Well Do the 100 Largest U.S. Banks Protect Their Domains? A DNS Security Study. https://www.webvello.com/research/bank-dns-security

Common questions

What is Webvello Research?

Webvello Research is the first-party research program of Webvello, a web design and AI search optimization company in Roseville, California. It publishes dated studies of how websites present themselves to search engines and AI assistants, with the method, raw counts and limits on each page.

Can I quote or reuse these findings?

Yes. Quote any figure with a link to the study it comes from. Published datasets are free to reuse under CC BY 4.0 with credit to “Webvello Research”.

How often are studies updated?

Each study is a snapshot dated to the day its data was collected. When we rerun one, we publish the new date and keep the earlier figures visible so changes over time can be compared.

Can you run these checks on my website?

Yes. The free AI visibility audit runs the same checks on any public domain: robots.txt rules per AI agent, CDN blocks, structured data and how the page reads to an assistant.

See how your site measures up

The free audit runs the same checks as these studies on your domain and sends a written plan of what to fix first.