Why Structured Data Matters for WordPress SEO
- Automatic structured data (JSON-LD) output — How Article schema affects Discover visibility
- Why structured data is necessary
- Why publication and modification dates are directly connected to E-E-A-T
- One featured image was not enough
- Do not invent copyright information
- The challenge of accurately identifying attachments
- Why robots meta needs to be enforced
- Security was part of the design as well
- Conclusion: SOHO SEO structured data
Automatic structured data (JSON-LD) output — How Article schema affects Discover visibility
So far, I’ve focused on the content quality assessment group.
Now we’re moving on to the indexing management group. The first topic is Article structured data—in other words, JSON-LD.
Earlier, I mentioned that the editorial checklist includes an item for checking whether structured data is being output. This article takes a closer look at what that check actually does behind the scenes.

Why structured data is necessary
When people read an article, they look at the headline, view the photos, and check who wrote it.
They can take all of that in at a glance and understand it naturally. Google’s crawlers, however, don’t read the way people do.
They have to parse the HTML and infer which part is the headline, which part is the body, and so on. Inference can go wrong.
I think of it as the difference between a face and a passport.
Guessing someone’s identity from their face is not as reliable as checking the information printed in their passport.
Article JSON-LD is a passport for an article. headline, image, datePublished, dateModified, author, and publisher: these six fields are the core information that SOHO SEO fills in automatically.
Why publication and modification dates are directly connected to E-E-A-T
I consider datePublished and dateModified especially important.
When Google decides whether to promote content in Discover, freshness carries significant weight. These two fields provide the basis for that assessment.
They give Google a way to distinguish an article that has been left with outdated information from one that has been updated recently.
The same applies to the author field.
It doesn’t just contain a name; it also sets @type to Person.
That is a structured declaration that the article was written by a real person. I use the default set by the administrator in the settings screen, but if the author’s account has a display name, I prioritize that instead.
The publisher field contains information about the organization. It represents the site’s own credibility.
If a logo image has been configured, I include it in the publisher object as the logo field.
Put these three elements together—the two date fields and the author and publisher information—and you have the framework of E-E-A-T.
Experience and expertise are expressed through the content itself, but authority and trustworthiness are supported by the minimum evidence that this structured data conveys to Google.
One featured image was not enough
This is where I put the most thought into the feature.
In the first version, I put only the featured image in the image field. But after looking through real users’ sites, I realized that was the wrong approach.
The photos people actually care about—the travel and restaurant photos they took themselves and deliberately kept with their copyright information—were usually images inserted in the body, not the featured image.
If those images are not exposed through the structured data at all, Google sees the article as if it contains only one photo.
The remaining images are not properly associated with the article in image search.

So now I collect both the featured image and the content images in the article and populate the field as an array.
When there are multiple images, I use an array; when there is only one, I use a single object. I don’t add images without limit, however.
Even competing plugins such as Yoast have no documented standard value for this, so I chose a practical limit that is generous enough for most cases.
Do not invent copyright information
For each image, I fill in the copyrightNotice and creator and creditText fields. I take this information from the EXIF and IPTC metadata that users intentionally left attached when they uploaded the photo.
WordPress reads those values as attachment metadata at upload time and stores them, so I use the values it has already saved.
I applied this principle consistently throughout the plugin.
If a value doesn’t exist, I leave it empty. I don’t force in an empty string or the site name just to fill the field.
Why does that matter?
I handled the license field according to the same principle. Google Images’ “Licensable” badge appears only when a license field is present.
If an administrator enters the URL of a copyright or usage-policy page in the settings screen, this field can be populated.
But I attach it only to images for which the copyright or creator information has actually been verified.
There’s a reason for that. With free public-domain images, adding it could appear to claim ownership of photos that do not belong to the user.
I didn’t want to create that kind of ambiguity. I chose to prioritize the principle of never inventing unsupported information over convenience.

The challenge of accurately identifying attachments
There was one more technically tricky issue.
I discovered that tracing an embedded image back to its attachment using only the image URL does not always work.
In some real cases, the image address recorded in the post pointed to an old filename ending in .jpg, while the actual attachment had since been converted to WebP and registered under a different filename.
But the img tag itself already contains a marker that WordPress adds automatically, such as class="wp-image-1400".
I changed the logic to prioritize the attachment ID read directly from this class instead of relying on URL tracing.
It is more accurate, requires no database lookup, and is therefore faster.
Only when the class is missing do I fall back to tracing the URL.
Fixing this small bug taught me that making good use of information WordPress has already generated is often better than building new inference logic from scratch.

Why robots meta needs to be enforced
This class does one more job besides supporting the structured data.
It ensures that max-image-preview:large is always present in the robots meta tag.
I’ve seen this kind of conflict several times.
That’s why I made SOHO SEO explicitly guarantee this value one more time at the end, regardless of whether other code has already modified it.
Without it, even if you follow the earlier checklist and use a featured image at least 1,200 pixels wide, the page may no longer qualify for display as a large image card.
Security was part of the design as well
There is one final point I want to mention.
When outputting JSON-LD, I use the JSON_HEX_TAG option.
This addresses a potential XSS vulnerability. Escaping < and > prevents those characters from appearing in the resulting string, shutting down the risk at its source.
This was an issue identified and fixed during the review process for the official WordPress directory.
Even when building a single SEO feature, I believe you have to take care of safeguards like these in the background.
Conclusion: SOHO SEO structured data
Article JSON-LD is not just another tag.
It is a structured way to show Google the article’s freshness through its publication and modification dates, its credibility through the author and publisher, and the authenticity of its content through image arrays and copyright information.
While building this, I followed two principles: never fill in unsupported information, and make the most accurate possible use of information the user has already provided.
In the next article, I’ll take a deeper look at the condition that works alongside this structured data: why max-image-preview:large is essential for appearing in image cards.

This article was designed using DECOBLOCKS.