DEV Community

Cover image for A Brand Name Alone Doesn't Count in ChatGPT's Business Hub. Here's the JSON-LD and DNS Fix
Marc Mulvaney
Marc Mulvaney

Posted on Edited on AI-assisted

A Brand Name Alone Doesn't Count in ChatGPT's Business Hub. Here's the JSON-LD and DNS Fix

OpenAI's ChatGPT Business Hub lets a business claim its listing. The Insights panel shows mentions and website clicks for the past 28 days. OpenAI defines a mention narrowly. For an online business, a mention counts only when ChatGPT references the business's website. The brand name alone does not count. For a local business, a mention counts when ChatGPT shows a result for that location.

The developer's job follows from that definition. Every crawler and every model must see one name tied to one URL. This post covers three tasks: pick one canonical name and URL, publish Organization or LocalBusiness JSON-LD, and claim the domain with a DNS TXT record. The checks and the common mistakes come after.

1. Pick one name and one URL

Write down two strings before you touch code.

  • The exact name. "Example Corporation" goes everywhere. "Example Corp." on LinkedIn and "ExampleCorp" on GitHub are two extra entities to a parser.
  • The exact home URL. Choose https. Choose www or non-www. Choose a trailing slash or none. Example: https://www.example.com/.

The name string goes in five places: the JSON-LD name, the home page <title>, og:site_name, the footer, and the name field of every profile you control. The URL goes in the website field of every profile. Every other host and scheme redirects to the canonical URL with a 301.

Check the redirects from a terminal:

for u in http://example.com/ https://example.com/ http://www.example.com/; do
  echo "$u"; curl -sI "$u" | grep -i -E "^(HTTP|location)"
done
Enter fullscreen mode Exit fullscreen mode

Each non-canonical variant returns HTTP/1.1 301 and a location: header that points at the canonical URL. A 302 or a 200 on a variant means the redirect is missing or temporary.

2. Organization JSON-LD on the home page

Google's Organization documentation places the markup on the home page, or on one page that describes the organisation. Google lists no required properties for Organization. Google recommends name, url, logo, sameAs, address, telephone, email and description. Google uses url to identify the organisation, so url holds the canonical URL from step 1.

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Organization",
  "@id": "https://www.example.com/#organization",
  "name": "Example Corporation",
  "legalName": "Example Corporation Inc.",
  "url": "https://www.example.com/",
  "logo": "https://www.example.com/images/logo.png",
  "description": "Example Corporation makes replacement parts for small manufacturers.",
  "email": "contact@example.com",
  "telephone": "+1-555-0100",
  "sameAs": [
    "https://www.linkedin.com/company/example-corporation",
    "https://www.youtube.com/@examplecorp",
    "https://github.com/examplecorp"
  ],
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "100 Main St",
    "addressLocality": "Springfield",
    "addressRegion": "IL",
    "postalCode": "62701",
    "addressCountry": "US"
  }
}
</script>
Enter fullscreen mode Exit fullscreen mode

Field rules from Google's documentation:

  • sameAs lists pages on other sites about the same organisation. List only profiles that show the exact same name and link to the exact same URL.
  • logo must be at least 112x112 px. The image URL must be crawlable and indexable. The file format must be one Google Images supports. The logo must look right on a purely white background.
  • legalName holds the registered name when it differs from the trading name. The trading name stays in name.
  • @id gives the node a stable identifier. A WebSite or WebPage node on another page references "publisher": {"@id": "https://www.example.com/#organization"} instead of repeating the block.

3. Local business: use the most specific LocalBusiness subtype

Google's local business documentation requires name and address. Google recommends geo, openingHoursSpecification, priceRange, telephone and url. Pick the most specific schema.org subtype: Plumber, Electrician, Dentist, HairSalon, Restaurant. A business with several trades lists them as an array, for example "@type": ["Electrician", "Plumber", "Locksmith"].

<script type="application/ld+json">
{
  "@context": "https://schema.org",
  "@type": "Plumber",
  "@id": "https://www.example.com/#business",
  "name": "Example Plumbing",
  "url": "https://www.example.com/",
  "image": "https://www.example.com/images/van.jpg",
  "telephone": "+1-555-0101",
  "priceRange": "$$",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "100 Main St",
    "addressLocality": "Springfield",
    "addressRegion": "IL",
    "postalCode": "62701",
    "addressCountry": "US"
  },
  "geo": {
    "@type": "GeoCoordinates",
    "latitude": 39.78173,
    "longitude": -89.65015
  },
  "openingHoursSpecification": [
    {
      "@type": "OpeningHoursSpecification",
      "dayOfWeek": ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday"],
      "opens": "08:00",
      "closes": "17:00"
    }
  ],
  "sameAs": [
    "https://www.yelp.com/biz/example-plumbing-springfield",
    "https://www.facebook.com/exampleplumbing"
  ]
}
</script>
Enter fullscreen mode Exit fullscreen mode

Rules for this block:

  • One entity gets one block. A local business publishes the subtype block, not a second Organization block with different data for the same business.
  • geo.latitude and geo.longitude need at least five decimal places.
  • opens and closes use 24-hour HH:MM times. A business open 24 hours sets opens to "00:00" and closes to "23:59". A closed day sets both to "00:00".
  • telephone includes the country code and the area code.
  • priceRange takes a currency-symbol string such as "$$" or a specific range such as "$100-$300".

4. Claim the domain with a DNS TXT record

The Business Hub offers the verification methods available for your business, such as business phone, business email, website domain via a DNS TXT record, or Shopify. After that step, OpenAI runs an identity check through Persona before the account can publish changes or view insights. The DNS route proves control of the domain itself and does not depend on anyone's inbox.

  1. In the hub, choose the website domain option. Copy the token.
  2. In your DNS provider (Cloudflare, Route 53, your registrar), add a TXT record. Use the exact record name the hub shows. Paste the token as the value. Set the TTL to 300.
  3. Wait five minutes. Query a public resolver from outside your network:
# use the exact record name the hub gave you
dig +short TXT example.com @1.1.1.1

# Windows
nslookup -type=TXT example.com 1.1.1.1
Enter fullscreen mode Exit fullscreen mode
  1. The output shows the token in quotes. Click verify in the hub.
  2. Leave the record in place. OpenAI does not say whether it rechecks the record, so keeping it costs nothing and keeps proof of control.

5. Check it worked

  • Rich Results Test at https://search.google.com/test/rich-results . Paste the home URL. The result lists one Organization or LocalBusiness item with zero errors.
  • Schema Markup Validator at https://validator.schema.org/ . It validates every schema.org type, including types Google shows no rich result for.
  • A script for a list of client sites. It pulls every JSON-LD block from a page and prints type, name, url and the sameAs count:
import json, re, sys, urllib.request

url = sys.argv[1]
req = urllib.request.Request(url, headers={"User-Agent": "jsonld-check/1.0"})
html = urllib.request.urlopen(req).read().decode("utf-8", "ignore")
pattern = r'<script[^>]+type=["\']application/ld\+json["\'][^>]*>(.*?)</script>'
blocks = re.findall(pattern, html, re.S | re.I)
print(f"{len(blocks)} JSON-LD block(s) found")
for raw in blocks:
    data = json.loads(raw)
    nodes = data if isinstance(data, list) else data.get("@graph", [data])
    for n in nodes:
        print(n.get("@type"), "|", n.get("name"), "|", n.get("url"), "|",
              len(n.get("sameAs", [])), "sameAs")
Enter fullscreen mode Exit fullscreen mode

Run it as python3 check_jsonld.py https://www.example.com/. A correct page prints exactly one organisation-type node, the exact name string and the exact canonical URL. A json.loads exception means the block has a syntax error and no parser reads it.

  • In the hub, Insights covers the past 28 days. Read the trend after a full month.

6. Six mistakes and their fixes

  1. Name without URL. Directory listings, podcast show notes, conference bios and "our partners" pages show the name and omit the link. Fix: add the canonical URL next to the name, then add those pages to sameAs.
  2. Mismatched names. The legal entity appears in one place and the trading name in another. Fix: name holds the trading name, legalName holds the legal one.
  3. url that differs from the live page. http instead of https, or the non-www host. Fix: copy the canonical URL from the browser address bar after the redirects settle.
  4. A logo nobody can fetch. robots.txt blocks it, a data URI holds it, or a login protects it. Fix: a plain public image URL.
  5. Two blocks that disagree. A theme or plugin injects an Organization block with the site title and you add another by hand. Fix: run the script above and remove one block when it prints two organisation nodes.
  6. A TXT record deleted after verification. Fix: leave the record and document it in the DNS change log.

A prompt to audit your profiles

Copy the name and website fields from each profile into a list. Paste this into the AI assistant you use:

My company's exact name is "[Example Corporation]" and its canonical website URL is "[https://www.example.com/]". Below is a list of my public profiles, each with the name and website value I copied from that profile:

[profile URL 1] | name shown: [...] | website shown: [...]
[profile URL 2] | name shown: [...] | website shown: [...]

For each line, output a table with: profile URL, name matches exactly (yes/no), website matches exactly (yes/no), the exact text I should change it to. Then list which profile URLs are safe to add to my sameAs array.
Enter fullscreen mode Exit fullscreen mode

Expectations

OpenAI states that claiming a business does not guarantee increased visibility. OpenAI states that advertising does not influence its independent answers. Local stores and restaurants can currently claim only an existing U.S. listing. The markup above removes the ambiguity between a name and a URL. OpenAI does not say that structured data affects mentions. The markup gives every parser the same name and URL pair to read.

Sources: OpenAI Help, "Setting up your business on ChatGPT" (https://help.openai.com/en/articles/20001635-setting-up-your-business-on-chatgpt). Google Search Central, Organization structured data (https://developers.google.com/search/docs/appearance/structured-data/organization) and Local business structured data (https://developers.google.com/search/docs/appearance/structured-data/local-business).

Marc Mulvaney, founder of freewebsite247, has built websites and done SEO since 1997. freewebsite247 builds websites for US local businesses with this markup included. Every site is built by humans, assisted with AI. The website is free, you just pay for the hosting.

If that's too much work, contact us and we'll make you a free website.

Top comments (1)

Collapse
 
citedy profile image
Dmitry Sergeev •

The user wants a comment for the video. Must follow developer instructions: short, casual, start lower case, specific reaction or question about the video. No marketing, no URLs. Must be a comment text only. We need to produce a comment that mentions something specific: maybe about JSON-LD, DNS fix, brand name. Could ask about how to verify business listing or about the impact on SEO. Make it one or two sentences, maybe fragment. Lowercase start. e.g., "anyone know if the DNS fix works