The previous articles all covered how to write the things “people read”: titles, descriptions, and how pages divide the work. Today we look at something written specifically for machines: structured data.
What does it do? In one sentence: it states plainly on the page, in a format machines can read directly, exactly who our company is.
Why it matters more in the AI era: because for AI to introduce you, it first has to pin down a set of facts — the full legal name, the year founded, where you are, what you do, and what credentials you hold. If that information is scattered across the body text, machines have to guess by “interpreting” it; structured data simply “tells” them, with no guessing required.
First, the effect: it fixes “saying it wrong” and “leaving it out”
We have run into two kinds of real problems before: AI wrote the company name with a homophone typo; and AI said it “couldn’t find any credential information for this company” — even though the credentials are published on the official website.
What these two problems have in common is this: the information is on the page, but the machine failed to extract it accurately. Structured data targets exactly this — it doesn’t change what you say, it just says it again in a way machines won’t misread.
The first item to get right: corporate identity information
Structured data can mark up many things (articles, products, Q&A, breadcrumbs…), but for most corporate websites, the highest-return first item is describing the company itself clearly. We recommend including at least the following:
| What to include | Why it matters |
|---|---|
| Legal full name | Matches your business license. This is the key to machines matching you against business registration records |
| Common short name / aliases | Customers usually say the short name; without it, machines may not recognize it as the same company |
| Founding date | One of the details AI almost always mentions when introducing a company |
| Address, phone | The direct basis for local search and “who should I contact” questions |
| Scope of business (list it out, don’t write it as a paragraph) |
Determines whether AI thinks of you when “recommending companies that do X” |
| One-sentence summary | Compresses founding year, location, core business, credentials, and scale into a single sentence — AI often quotes the whole sentence verbatim |
One important line to draw: write your aliases, but never write a “wrong name” into them.
When we found AI writing the company name with a homophone typo, the thought crossed our mind to “just add the wrong spelling to the aliases too, so machines know it means us.” That is wrong. Structured data is a formal declaration you make to machines — putting a wrong name into “aliases” is the same as admitting out loud that “we’re also called that,” which only reinforces the error.
The right approach: put only the correct short names and English names in the aliases; as for common misspellings, address them in the body text as a correct-versus-incorrect comparison (for example, “The standard name of our company is X; a misspelling Y occasionally appears online — please take X as authoritative”). One is a declaration, the other is a clarification, and the two must not be mixed.
The three most common pitfalls (we’ve stepped in all of them)
Pitfall 1: you wrote it, but it never took effect.
This one is the most maddening. When auditing our own site, we found that several articles clearly contained a full block of structured data, but because it wasn’t properly wrapped in the tags it belonged in — it was neither recognized as structured data nor displayed on the page, a complete waste that sat in the database for months.
Common causes: the format got altered by the editor when pasting, it was placed in the wrong spot, or it was stripped out by a minification or optimization plugin. So “always verify after adding” is not optional.
Pitfall 2: it doesn’t match the page’s body text.
We fixed one of these just a couple of days ago: a new line of business was added to the website’s body text, but the structured data and page description were still on the old version and didn’t mention it.
The consequence: when AI introduces us, it has no idea that new line of business exists — because these structured fields are exactly what it trusts first. Whenever the body text changes significantly, remember to go back and check whether these “machine-facing” spots have kept up.
Pitfall 3: several copies on the same page, contradicting each other.
The theme ships with one, the SEO plugin generates one, and the web agency that built the site years ago added another by hand — and the three don’t even agree. When machines read contradictory information, they may simply disregard all of it.
What to do: first find out exactly how many copies are on the page (see the method below) and clean up the duplicates and outdated ones. If different sources each carry their own value, the professional approach is to make them “point to the same entity” — that way machines merge them into one company rather than treating them as several. We recommend leaving this step to someone who knows what they’re doing.
How to verify: two methods, five minutes
- Use the official testing tools. All the major search engines offer free structured data / rich results testing tools — paste in the URL and you can see what they read and whether any errors came up. Nothing read = nothing took effect, whether you wrote it or not.
- Look at the page source directly. Right-click the page, choose “View page source,” and search for the keyword to see whether that data is there and whether it’s complete. While you’re at it, count how many copies appear — that’s how you check for “Pitfall 3.”
The point of verification isn’t “I wrote it,” but “what the machine read.” These two things are often not the same.
Don’t expect it to work on its own
Let’s be honest: structured data is not a master key. It won’t suddenly get a company nobody has heard of recommended, and it can’t replace the content itself.
What it really does is reduce errors:
- Something that used to get your name wrong now gets it right;
- Something that used to miss one of your business lines now mentions it;
- Something that used to confuse you with another company now tells you apart.
None of these improvements looks big on its own, but they all happen at the exact moment AI is about to mention you — and at that moment, whether it gets you right or wrong makes all the difference.
So its proper role is this: foundational work done alongside your content. First make sure you get found (channels), then make sure you get understood (structured data), and only then get chosen (content and credentials). All three are essential, and the order can’t be reversed.
A Few Common Questions
Q: Does this require a programmer?
A: Most website builders’ SEO plugins already include basic business information settings — you just fill in a form, no coding required. But adding more detailed fields (such as aliases or a list of business scope) usually takes one round of technical help, after which you won’t need to touch it again.
Q: How long before it takes effect?
A: It has to wait for search engines to re-crawl, usually weeks to months. Remember to submit it through the webmaster tools after adding it — don’t just wait.
Q: Does every page need it?
A: No. For business information, one copy on the homepage and the company profile page is enough (or a single site-wide output). Article pages can separately mark up “this is an article, who wrote it, when it was published,” but that’s a lower priority than business information.
Q: Will adding it get me flagged as cheating?
A: As long as what you write is true, no. The rule is clear: structured data must match what people can actually see on the page. Putting in things that aren’t on the page (like fabricated ratings or credentials that don’t exist) is what counts as a violation, and you could be penalized for it.
Not sure whether the structured data on your website is actually working?
Siwenlide provides website and SEO/GEO content-driven lead generation operations for small and medium-sized businesses in Beijing and the surrounding area: checking how many pieces of structured data are actually on your pages, whether any contradict each other or have long since stopped working, and whether they match the body content, then filling in the missing fields in your business identity information and verifying that machines really do read it. The three pitfalls in this article are all ones we found and fixed on our own site. Visit siwenlide.com (or start with our company profile) or call 400-686-2011 to talk.
