Method

Five layers between a buyer’squestion and your name.

Five layers between a question and your name.

I read them top to bottom. Each one has a question, a place to look, and defects that are easy to see once you know where. You can check every one of them yourself.

Sources

Are you in the places the assistant reads?

Weinführer.de · listing

VerkostungVersand DEDeutsch

Weinführer.de · listingYour wineryIstrien, Kroatien

No description
No languages
No shipping

Same directory, same region. The assistant has three paragraphs to quote about them and one line about you.

What I read

Directories and guides in the buyer’s language
Review and comparison sites
The sources each answer cites

Typical defects

Listed, but as a bare name
Missing from the one guide every answer cites
Old address or closed season

Check it yourself

Ask an assistant your buyer’s question in their language. Open every source it names. Search each for your business.

Identity

Does the machine know who you are?

<script type="application/ld+json">
{
  "@type": "Winery",
  "name": "Home",
  "url": "https://your-winery.hr/de/",
  "inLanguage": "hr"
}
</script>

Should read

“name”: “Your Winery”

The template filled the name with the page title. To a machine, this business is called Home, and its German page says it is in Croatian.

Structured data from a German page. Two fields wrong, both invisible to a human visitor.

What I read

JSON-LD and meta tags on every template
Business profiles and knowledge panels
How your name is spelled across sources

Typical defects

The name field reads “Home”
Three spellings of one brand
Language tags that contradict the page

Check it yourself

Paste your German page into a rich results test. Read the name field out loud.

Language

Does your page sound like it was written for them?

Written in German

Unsere Weine reifen in Fässern aus istrischer Eiche.

Machine translated

Unsere Weine sind gealtert in Fässern von Eiche von Istrien.

Both say “our wines age in Istrian oak barrels”. A German reader, and a model trained on German, can tell which one is a translation.

What I read

Every page in each market language
Titles and descriptions as they show in results
The words buyers actually use

Typical defects

Machine-translated pages, unedited
German page, Croatian title tag
The product named in the seller’s terms, not the buyer’s

Check it yourself

Send your German page to a German friend. Ask one question: would you call?

Crawl

Can the machine reach the page at all?

your-winery.hr/de/weinkeller-tour200 OK

WillkommenThis is the homepage

Soft 404

The tour page was deleted. The server answers “fine” and serves the homepage instead.

So the old link still counts, but for the wrong page, and the real tour never gets read.

A missing page that pretends to exist. Should return 404, or redirect to the real tour.

What I read

Every URL a crawler can find
Status codes, redirects, canonicals
hreflang between market versions

Typical defects

Missing pages that return the homepage
Language versions pointing at each other wrongly
Pages that only render with scripts

Check it yourself

Type your domain plus /de/anything-made-up. If you see your homepage, you have this defect.

Demand

Do you answer what they actually ask?

“Weingut Istrien Verkostung”How often it is asked

Your pages that answer itHow well you rank

Asked versus ranked. The dashed length is demand you have no page for. Illustrative proportions.

What I read

What buyers search and ask, per market
Which of your pages could answer it
Who ranks instead, and with what

Typical defects

No page for the question buyers ask most
Two pages competing for one question
Pages written for words nobody types

Check it yourself

Write down the three questions your foreign buyers ask on the phone. Find the page that answers each one.

I read all five for you in two weeks.

Get a free audit