VideoObject Schema: creator Property and interactionStatistic

In short: On 24 September 2026, Google updated its VideoObject documentation to add the creator property (noting author is also supported) and to spell out which interaction types interactionStatistic accepts: WatchAction, LikeAction, CommentAction and ShareAction. Both are recommended, not required. Neither unlocks a new rich result. Both matter more than that suggests.

1. What Happened? VideoObject Schema: creator Property and interactionStatistic

On 24 September 2026, Google updated the VideoObject structured data documentation with two changes.

First, creator is now documented as a recommended property. It accepts a Person or an Organization, and describes the person or organisation that created or published the video. Google notes that author is also supported as an alternative. Two sub-properties are documented alongside it:

• creator.name (or author.name) – the name of the creator or organisation. If you specify a person or organisation, you must provide either name or alternateName.
• creator.url (or author.url) – a link to a web page that uniquely identifies the creator or organisation, such as a profile page or homepage.

Google’s own reference example on that page now carries a creator block.

Second, interactionStatistic now documents exactly which interaction types Google supports. Previously most implementations used WatchAction and assumed the rest were unsupported. Four types are now named explicitly:

Interaction type	What it counts
https://schema.org/WatchAction	Times the video has been watched
https://schema.org/LikeAction	Likes or upvotes
https://schema.org/CommentAction	Comments
https://schema.org/ShareAction	Reshares

Each is declared inside an InteractionCounter with a userInteractionCount integer.

What did not change. The required properties are still just three: name, thumbnailUrl and uploadDate. creator and interactionStatistic both sit in the recommended list. And Google’s standing caveat applies – it does not guarantee that features consuming structured data will appear in search results.

So this is a documentation clarification, not a feature launch. That is precisely why it is worth reading carefully: Google rarely documents a property unless something consumes it.

2. Why Does It Matter?

creator.url is an entity identifier, not a link

Read the wording again: a link to a web page that uniquely identifies the creator. That is not “a link to your about page.” That is disambiguation language – the same job sameAs does on Organization, and the same job a profile page does for an author entity.

Every video on your site currently sits as a fairly anonymous object. Adding creator with a stable, resolvable URL connects each one to an entity that Google can reconcile across your site, across the web, and across everything else that entity has produced. For an organisation publishing hundreds of videos under a handful of named experts, that is a meaningful consolidation of authorship signal that currently exists nowhere in the markup.

This is the strategic point, and it is easy to miss because creator looks like a trivial metadata field. It is entity plumbing, not a rich-result feature. In a search environment moving steadily towards entity-based retrieval and attributed answers, declaring who made a thing – with an identifier rather than a name string – is the kind of work that compounds.

Four interaction types opens up engagement data that most sites are throwing away

Almost every VideoObject implementation in the wild declares WatchAction and stops. The documentation now confirms that likes, comments and reshares are all supported types.

If you run a platform with genuine social interaction – a community, a course platform, a UGC video service, a publisher with active comment threads – you hold engagement data that has never been expressible in your markup and is now documented as supported. That is a real expansion of what you can say about a video.

But be clear-eyed about what it does not do

There is no documented video rich result that displays like, comment or share counts. Nothing in this update promises a visual change in the SERP. Any team briefing this work as “it’ll make our videos stand out in search results” is setting up a disappointment.

The honest case for implementing is different and better: it makes your data machine-readable and correct, which matters increasingly for systems that consume structured data outside the classic rich-result pipeline. Implement it because it is true and useful, not because it is promised to show.

The risk nobody will brief you on

Structured data that misrepresents the page is a guidelines violation, and Google’s documentation carries an explicit warning that markup outside its structured data guidelines can trigger a manual action.

Engagement counts are unusually tempting to inflate. Someone will suggest rounding 847 views up to “about 1,000”, or declaring a like count the page does not actually display. Do not. The general structured data guidelines require markup to represent the page’s actual content, and engagement counts are exactly the sort of claim that is trivially checkable against the rendered page.

There is a subtler version of the same risk: stale counts. Markup declaring 5,647,018 views, hardcoded eighteen months ago into a template, is now a false statement about your page. Engagement data is dynamic. If you cannot keep it current, do not declare it.

3. Who Is Affected?

Publishers and media. The highest-value use of creator as a Person. Journalists and presenters already have author profile pages; connecting every video to them is a small change with real entity benefit. Active comment threads make CommentAction genuinely available.

E-learning and education platforms. Instructor-led video is the norm, instructors have profile pages, and course platforms usually hold completion and engagement data. The creator as Person with a faculty or instructor profile URL is close to a perfect fit.

Healthcare and regulated sectors. Where a named clinician appears in a video, creator pointing at a credentials page connects the content to a verifiable professional. That is the same expertise signal these organisations already work hard to establish in text, expressed in a form machines can read.

E-commerce. Product demonstration, unboxing and how-to videos. creator as Organization pointing at your homepage or brand page ties video assets to the brand entity you are already building. Review and demo videos with genuine engagement data can carry it.

SaaS and B2B. Product walkthroughs, webinar recordings, release notes. Organization for brand-produced content, Person where a named developer advocate or founder presents.

Video platforms and UGC services. The largest beneficiaries of the interaction types. If you run a platform where users watch, like, comment and share, you can now declare all four, and you have the infrastructure to keep them current.

By team:

• Product owns whether engagement counts are exposed in the data layer at all, and whether creator is a first-class entity in the CMS.
• Content owns whether every video actually has an identified creator with a live profile page.
• Development owns markup generation, count freshness and validation.

4. What Should Businesses Do?

Step 1 – Decide your creator model (Product + Content – one week)

Before writing any markup, answer one question: is the creator of your videos a person or the organisation?

Pick per content type, not per video, and write it down:

Content type	Creator	Why
Journalism, expert explainers, instructor-led courses	Person	The individual is the expertise signal
Product demos, brand campaigns, corporate comms	Organization	No individual is the meaningful author
Webinars with named presenters	Person	Presenter carries the authority
Anonymous or team-produced content	Organization	Do not invent a person

Then make sure the URL you will point at actually exists. creator.url is supposed to uniquely identify the creator – a live author profile, an instructor page, a clinician’s credentials page, or your homepage for an organisation. If your named experts do not have profile pages, that is the prerequisite task, not the markup.

Step 2 – Implement creator (Development – one sprint)

Person, for a publisher or education platform:

json

{
  "@context": "https://schema.org",
  "@type": "VideoObject",
  "name": "How to read a cholesterol test result",
  "description": "A consultant cardiologist explains each value on a standard lipid panel and what the reference ranges mean.",
  "thumbnailUrl": ["https://example.com/thumbs/16x9/lipid-panel.jpg"],
  "uploadDate": "2026-09-14T09:00:00+01:00",
  "duration": "PT8M22S",
  "contentUrl": "https://example.com/video/lipid-panel.mp4",
  "embedUrl": "https://example.com/embed/lipid-panel",
  "creator": {
    "@type": "Person",
    "name": "Dr Amara Okonkwo",
    "url": "https://example.com/experts/amara-okonkwo"
  }
}

Organization, for e-commerce or SaaS:

json

{
  "@context": "https://schema.org",
  "@type": "VideoObject",
  "name": "Setting up the X200 espresso machine",
  "description": "Step-by-step first-time setup, from unboxing to your first shot.",
  "thumbnailUrl": ["https://example.com/thumbs/16x9/x200-setup.jpg"],
  "uploadDate": "2026-09-02T11:30:00+01:00",
  "duration": "PT4M05S",
  "contentUrl": "https://example.com/video/x200-setup.mp4",
  "creator": {
    "@type": "Organization",
    "name": "Northfield Coffee Co",
    "url": "https://example.com"
  }
}

Two implementation notes. Use creator or author, not both, and be consistent across the estate – mixed usage makes auditing miserable for no benefit. And if you specify a creator, you must give it a name (or alternateName); an empty or partial creator object is worse than omitting the property.

Step 3 – Implement interactionStatistic honestly (Development + Product – one sprint)

Declare only the interaction types your platform genuinely tracks and displays. Multiple counters go in an array, one per type:

json

{
  "@context": "https://schema.org",
  "@type": "VideoObject",
  "name": "Community build: restoring a 1972 lathe",
  "thumbnailUrl": ["https://example.com/thumbs/16x9/lathe-restore.jpg"],
  "uploadDate": "2026-08-19T18:00:00+01:00",
  "creator": {
    "@type": "Person",
    "name": "Tomas Lindqvist",
    "url": "https://example.com/members/tomas-lindqvist"
  },
  "interactionStatistic": [
    {
      "@type": "InteractionCounter",
      "interactionType": { "@type": "WatchAction" },
      "userInteractionCount": 48213
    },
    {
      "@type": "InteractionCounter",
      "interactionType": { "@type": "LikeAction" },
      "userInteractionCount": 3106
    },
    {
      "@type": "InteractionCounter",
      "interactionType": { "@type": "CommentAction" },
      "userInteractionCount": 274
    },
    {
      "@type": "InteractionCounter",
      "interactionType": { "@type": "ShareAction" },
      "userInteractionCount": 991
    }
  ]
}

The freshness rule. Engagement counts are dynamic data and must be generated at render time from the same source that populates the visible page. Three acceptable patterns:

1. Server-rendered – inject counts into the JSON-LD during page render from your live data layer. Cleanest option.
2. Cached with a short TTL – acceptable if the cache window is measured in hours, not months. Exact real-time parity is not required; being wrong by an order of magnitude is not acceptable.
3. Omit the property – entirely legitimate. If you cannot keep counts current, leave them out.

What is not acceptable is a hardcoded number in a template. Add a lint rule if you have to.

If you are still using the older interactionCount property, note that Google has recommended interactionStatistic since October 2019 and continues to support both. Migrate when you next touch the template; there is no urgency, but there is no reason to write new interactionCount markup either.

Step 4 – Validate and monitor (Development – ongoing)

• Run representative URLs through the Rich Results Test and fix critical errors. Non-critical warnings are worth fixing too – they usually indicate exactly the sort of partial implementation described above.
• Use the URL Inspection tool on live pages to confirm Google sees what you think it sees, particularly if your markup is generated client-side.
• Watch the Video rich result report in Search Console after deployment. The pattern to watch for is an increase in valid items with no increase in invalid ones. A jump in invalid items after a release almost always means a template regression.
• Add structured data validation to CI. A schema change that silently breaks markup across 4,000 video pages is a bad thing to discover in a quarterly audit.

Step 5 – Fix the prerequisite you probably have (Content – ongoing)

Most organisations will discover, on attempting Step 1, that their named experts do not have stable profile pages – or have three of them, at inconsistent URLs, with different biographies.

That is the real work this update surfaces. creator.url is only as useful as the page it points at. A profile page that states who the person is, what they are qualified in, and what else they have produced turns the property from a string into an entity. Without one, you are declaring a name.

5. What We’re Watching Next

• Whether engagement counts start surfacing anywhere. Google has documented four interaction types without announcing any feature that displays them. Documentation usually precedes consumption. Video carousels, Discover and AI-generated summaries are all plausible destinations for social-proof signals.
• Whether creator becomes a stronger attribution signal in AI surfaces. As generated answers increasingly cite and attribute, a machine-readable statement of who made a piece of content – with a resolvable identifier – is exactly the input an attribution system needs. We would expect creator-level entity data to matter more over the next year, not less.
• Whether Google tightens on inflated counts. Engagement numbers are easier to verify against a rendered page than most structured data claims. If declared counts become widely gamed, this is an obvious enforcement target.
• Whether creator spreads to other types. The property exists across much of schema.org. Its documentation here, in a type Google actively maintains, may signal a broader push towards explicit creator attribution across content types.

About Szymaniak Digital

Szymaniak Digital is an enterprise AI SEO consultancy working with brands on visibility across classical search and AI-mediated discovery.

Documentation changes like this one get filed as “minor” and skipped. The pattern we see repeatedly is that the small, unglamorous entity work – who made this, what is it connected to, is the data current – compounds quietly while teams wait for a feature announcement that justifies the effort.

If you want your video structured data audited across the site, or your creator entity model built properly before the markup goes in, that is work we do.

Contact Us!