Sample report

Sample report. Your report will look the same. This is a real check of moregroup.estate, a Phuket property site we run, as it came out on 28 September 2026. Nothing here was edited.

This site already has an llms.txt and lets AI crawlers in, so those two checks show as passed. On a site without them, the report drafts the llms.txt from your own pages and, if your firewall turns crawlers away, starts with the setting to change.

89 OF 100

AI visibility check · 28 September 2026 · 0.7 s

moregroup.estate

AI visibility A 89/100 SEO basics A 95/100

AI assistants can read and quote you. The open question is whether they name you.

Buyers ask ChatGPT who to buy from. It sends them to 5 other businesses.

We asked ChatGPT "best property advisors for foreign buyers in Phuket". It named Subata Realty, Butler Estates, Phu Phrom Property and 2 more. moregroup.estate was not one of them.

See every week who ChatGPT names

7 days free. Cancel any time.

Who ChatGPT names instead of you

3 buyer questions

ChatGPT named moregroup.estate in 0 of 3 answers.

We asked it the way a buyer looking for Phuket property shortlist advisory would.

  • "best property advisors for foreign buyers in Phuket"

    not named Named instead: Subata Realty, Butler Estates, Phu Phrom Property, Easy Living Phuket, Neo Realty Phuket

    Not counted as competitors: Walailak Law Firm, Siam Legal Phuket, Khonsu Legal

    Also mentioned: Costa Global Capital, Storm Phuket

  • "who can help me shortlist Phuket condos and villas based on budget and preferred use"

    not named Named instead: Pearl Property Phuket, MechThai, InDreams Phuket Real Estate, ZEN Real Property, WAVE

    Not counted as competitors: FazWaz, DDproperty, Hipflat

  • "top real estate agencies or buyer’s agents to consult for Phuket pre-sales and off-plan properties"

    not named Named instead: Anan Property Group, Launay Properties, Home In Phuket, Nova Property

    Not counted as competitors: Walailak Law Firm, GoLaw Phuket

    Also mentioned: OSTRO, Pansea Brokers, Storm Phuket, Costa Global Capital

That was three questions, asked once. Every week Watch asks ChatGPT and Perplexity ten questions like these, names your rivals and sends a paragraph for your site so the next answer names you.

OperStack WatchMonday, 5 October

moregroup.estate: named in … of 10 answers.

Named instead of you: Subata Realty.

This week's fix: a paragraph for your homepage.

What the email looks like. Your numbers arrive after the first check.
Send me this every week

7 days free. Cancel any time from the link in every email. Then $35 a month.

Answers come from ChatGPT through its API with web search switched on, asked on 28 September 2026. They change from run to run, so treat one run as a sample, not a verdict.

In plain words

AI assistants can read your site and have something to quote. What decides whether they name you now is mostly what other sites say about you.

Your pages do not open with a short, direct answer. That opening paragraph is what assistants quote.

What to do first

  1. youDo the first fix. It is written out below, with how to check it worked.
  2. youThen work through the other finding: each is written out in Watch or the one-off audit.

For your developer

1 item

First fix: Is there something to quote

Pages: /, /gcc, /news, /areas, /terms

  • 5 of 10 measured page(s) open with an answer-first paragraph (20 to 90 words with a figure right after the H1). That paragraph is what gets quoted. Missing on: /, /gcc, /news, /areas, /terms.
  • Task: Rewrite the opening so the first paragraph answers the question the page was opened for: twenty to ninety words, carrying a concrete figure (a price, a timing, a share, a count). Get the figure from the site owner and name its source beside it. If there is no figure, do not invent one: ask the owner and leave a marker.
  • Check: The first paragraph after the heading is 20 to 90 words and carries a figure with a named source.

Readiness by engine

5 engines

An engine’s score uses the same formula as the overall score, but its access counts only the crawlers that engine reads pages with for answers (named under each engine). So on a site that blocks nobody the numbers match the overall score.

ChatGPT

89

ChatGPT search (OAI-SearchBot), ChatGPT browsing (ChatGPT-User)

Perplexity

89

Perplexity (PerplexityBot), Perplexity browsing (Perplexity-User)

Claude

89

Claude search (Claude-SearchBot), Claude browsing (Claude-User)

Gemini

89

Gemini grounding (Google-Extended)

Copilot

89

Copilot and Bing (Bingbot)

What else can be improved

5 areas

8 checks passed, 1 thing to fix.

Can AI crawlers read it worth 25 25/25
  • passAll 14 AI crawlers and fetchers are allowed (Content-Signal: search=yes, ai-input=yes, ai-train=yes).
Is there a map for agents (llms.txt) worth 15 15/15
  • passllms.txt present with 1372 link(s) to the site's own pages.
Is the entity clear (schema) worth 20 20/20
  • passOrganization or business schema names the entity behind the site.
  • passFAQPage schema found, the easiest format for an engine to quote.
  • passService schema on the sampled pages: an engine can name what it is quoting.
Is there something to quote worth 25 19/25
  • passSampled pages carry 1533 words on average.
  • passMeasured on these pages: /, /gcc, /usa, /book, /news, /team, /about, /areas, /india, /terms. Every content finding is a share of them, so a fix on one page moves the score by its share.
Can it be dated and trusted worth 15 10/15
  • passSampled pages expose publication dates.
  • to fix1 thing in this area, listed below

SEO basics on the same pages

95/100

Worked out from what was already read, with no extra requests. These are the basics, not a full audit: speed, links and rankings are not here.

  • 10/10Title of 10 to 70 characters
  • 7/10Description of 50 to 170 characters
  • 10/10An H1 on the page
  • 10/10A canonical address
  • 10/10An og:title for link previews
  • 10/10Not closed to indexing

The one change to make first

Is there something to quote
Now
5 of 10 measured page(s) open with an answer-first paragraph (20 to 90 words with a figure right after the H1). That paragraph is what gets quoted. Missing on: /, /gcc, /news, /areas, /terms.
Task
Rewrite the opening so the first paragraph answers the question the page was opened for: twenty to ninety words, carrying a concrete figure (a price, a timing, a share, a count). Get the figure from the site owner and name its source beside it. If there is no figure, do not invent one: ask the owner and leave a marker.
How to check
The first paragraph after the heading is 20 to 90 words and carries a figure with a named source.

Hand this to whoever runs the site. It is written so it can be done without asking us anything.

A paragraph ChatGPT can quote, written from your own site

your homepage

Add this right below the H1 of your homepage, as its first paragraph.

MORE Group helps buyers compare Phuket property with a three-project shortlist matched to budget and use. We match construction status and buyer checks before reservation, and each project page shows location and status. Featured projects include Origin Residences Bang Tao from $133,000, So Origin Bangtao Beach from $149,000, and Essence Residence Rawai from $148,000. We also offer flexible payment plans, a 0% interest construction payment plan, and up to 15 years Thai bank mortgage.

Optional: if you offer a shortlist starting from $80K, add this sentence: Shortlists start from $80K, and featured options include Origin Residences Bang Tao, So Origin Bangtao Beach, and Essence Residence Rawai.

This paragraph answers one question. Watch writes a new one every week, for the questions where rivals beat you now, and shows a week later whether it worked.

Get a paragraph like this every week

7 days free. Cancel any time from the link in every email. Then $35 a month.

One more reason buyers may not hear your name

written out in Watch
  • partial Can it be dated and trusted: 9 of 9 sampled

The list is real and this is its length. Watch writes out every line the same way as the fix above, with the text to paste, and checks every week whether ChatGPT and Perplexity name you. Full fix list and weekly progress unlock with your first payment.

Start 7 days free

7 days free. Cancel any time from the link in every email. Then $35 a month.

Or the full one-off audit, $149.

The pages we read

10 of 1430 in the sitemap
PageWordsH2TablesLinksImagesAnswer paragraph with a figureDate in markupSchema
/ 1858 18 1 77 10 no yes RealEstateAgent, Person, QuantitativeValue, Place
/gcc/ 665 5 1 19 2 no yes RealEstateAgent, Person, QuantitativeValue, Place
/usa/ 613 4 0 14 2 yes yes RealEstateAgent, Person, QuantitativeValue, Place
/book/ 466 4 0 4 2 yes yes RealEstateAgent, Person, QuantitativeValue, Place
/news/ 6739 191 0 382 2 no yes RealEstateAgent, Person, QuantitativeValue, Place
/team/ 720 6 0 3 3 yes yes RealEstateAgent, Person, QuantitativeValue, Place
/about/ 1027 10 0 21 2 yes yes RealEstateAgent, Person, QuantitativeValue, Place
/areas/ 1110 10 1 45 7 no yes RealEstateAgent, Person, QuantitativeValue, Place
/india/ 1563 12 1 21 2 yes yes RealEstateAgent, Person, QuantitativeValue, Place
/terms/ 564 10 0 4 2 no yes RealEstateAgent, Person, QuantitativeValue, Place

Other checks read one page. This one read 10.

The crawlers we asked about

14, none blocked
CrawlerWhat it doesrobots.txt
ChatGPT search (OAI-SearchBot) fetches pages to cite them let in
ChatGPT browsing (ChatGPT-User) fetches pages to cite them let in
OpenAI training (GPTBot) collects text for training let in
Perplexity (PerplexityBot) fetches pages to cite them let in
Perplexity browsing (Perplexity-User) fetches pages to cite them let in
Claude (ClaudeBot) collects text for training let in
Claude search (Claude-SearchBot) fetches pages to cite them let in
Claude browsing (Claude-User) fetches pages to cite them let in
Anthropic (anthropic-ai) collects text for training let in
Gemini grounding (Google-Extended) collects text for training let in
Copilot and Bing (Bingbot) fetches pages to cite them let in
Apple Intelligence (Applebot-Extended) collects text for training let in
Common Crawl (CCBot) collects text for training let in
DuckDuckGo AI (DuckAssistBot) fetches pages to cite them let in

Now check your own site

About a minute, free, no signup. Your report opens at its own address, the same as this one.