On March 24, 2026, Google added more supported properties to its Discussion Forum and Q&A Page structured data documentation. If your site runs forum threads, support discussions or user-submitted questions, your existing valid markup keeps working.
The first job is to decide which of the new properties describe what your pages visibly show. Two are worth looking at first, commentCount and digitalSourceType. Google lists both as recommended for its feature, not required, so each belongs only on a page where it matches the visible thread.
Start by naming each template as a general thread or a single-question page, then add what fits. After deployment, test the live URL in Google's Rich Results Test and watch the reports in Search Console.
What Google changed on March 24, 2026
The change appears in the Google Search documentation updates log. The March 24, 2026 entry reads: "Added more supported properties for Discussion Forum and QA Page markup." Google gives the reason as "For providing more clarity on comment thread structure to Google ingestion systems. This prevents misinterpretations in our handling of forum and Q&A content."
Google wants a clearer picture of who said what in a thread, how many replies exist, and which replies came from a person. The update touches two feature pages, Discussion Forum structured data and Q&A Page structured data.
The sites affected are the ones with user-generated discussion threads, community forums, support discussions and Q&A pages. Existing valid markup remains usable. The update changed what Google documents as supported, not the baseline of required properties, so a page that was eligible before the update does not become ineligible because a new property is missing.
The new properties and what each one describes
Google's feature pages cover five additions or expansions. The first is digitalSourceType, which says whether a piece of content was created by a trained model, such as a large language model, or by a simpler automated process, such as a reply bot. Google supports two values. TrainedAlgorithmicMediaDigitalSource is for content from a trained model. AlgorithmicMediaDigitalSource is for content from a simpler algorithmic process. The Discussion Forum page says that when the property is omitted, Google assumes the content is human-generated.
The second is commentCount, the total number of comments, including comments you leave out of the markup. That makes it useful on long threads where the JSON-LD lists only some replies. On a Q&A page, Google says answerCount plus commentCount should equal the total number of replies of any type.
The third is comment, which holds clarifications and threaded replies that are not themselves answers. The fourth is the nesting of Comment objects, which should follow the parent-child relationships visitors can see on the page. The fifth is sharedContent on DiscussionForumPosting, which now covers WebPage, ImageObject, VideoObject, DiscussionForumPosting and Comment. Use it when a post mainly shares a link, an image, a video, or a quoted post or comment.
None of these is required for eligibility. For Discussion Forum, Google still requires author, author.name, datePublished, and at least one of text, image or video, unless the post is represented on another page with an external url. For Q&A, it requires mainEntity, one Question, answerCount, name, and at least one accepted or suggested answer for rich result eligibility. An Answer and a Comment each need text. These are the requirements for these Google features, not for Schema.org as a whole.
Choose DiscussionForumPosting or QAPage first
Before adding any property, settle which type a template uses. Use DiscussionForumPosting for a general discussion, where people share experiences, opinions, media or advice, and a thread may run from an opening post through nested replies. Use QAPage when the page is built around one user-submitted question and its answers, and visitors can offer alternative answers.
Google says a site should not use both types on one page just to increase coverage. It also says QAPage is not for a publisher-written FAQ page that has no user-submitted alternative answers. A gardening forum, for example, would mark a long "what are you growing this spring" thread as a DiscussionForumPosting, and a page titled "Why are my tomato leaves curling?" with a ranked set of answers as a QAPage.

A comment that clarifies the question, such as "Does the rattling happen at every speed?", is a Comment, not an Answer. An answer proposes a solution. Keeping that line clean matters because answerCount and commentCount count different things.
If you are still deciding which types belong on a site at all, the guide to types of schema markup covers the common ones, and the JSON-LD schema examples show complete code for several of them.
Discussion forum markup before and after
This is valid baseline markup for a human-written forum thread. It has the required properties plus url, headline and one comment. Every name, URL and sentence is fictional, and the URLs use the reserved example.com domain. Replace them with the values your page visibly shows.
{
"@context": "https://schema.org",
"@type": "DiscussionForumPosting",
"@id": "https://example.com/forum/blue-lanterns#post",
"url": "https://example.com/forum/blue-lanterns",
"headline": "How do I make a blue lantern brighter?",
"text": "I built a small blue lantern for my campsite, but the light is too dim. What should I check first?",
"author": {
"@type": "Person",
"name": "Mira Chen",
"url": "https://example.com/profiles/mira-chen"
},
"datePublished": "2026-03-20T14:30:00Z",
"comment": [
{
"@type": "Comment",
"text": "Check whether the battery is fully charged and whether the diffuser is blocking too much light.",
"author": {
"@type": "Person",
"name": "Noah Reed",
"url": "https://example.com/profiles/noah-reed"
},
"datePublished": "2026-03-20T16:00:00Z"
}
]
}The next version adds commentCount and a second comment from an automated helper. The count is 2 because the thread shows two comments. The human comment stays unlabeled, since Google assumes human-generated content when digitalSourceType is absent. The bot comment carries AlgorithmicMediaDigitalSource because a simple reply bot wrote it.
{
"@context": "https://schema.org",
"@type": "DiscussionForumPosting",
"@id": "https://example.com/forum/blue-lanterns#post",
"url": "https://example.com/forum/blue-lanterns",
"headline": "How do I make a blue lantern brighter?",
"text": "I built a small blue lantern for my campsite, but the light is too dim. What should I check first?",
"author": {
"@type": "Person",
"name": "Mira Chen",
"url": "https://example.com/profiles/mira-chen"
},
"datePublished": "2026-03-20T14:30:00Z",
"commentCount": 2,
"comment": [
{
"@type": "Comment",
"text": "Check whether the battery is fully charged and whether the diffuser is blocking too much light.",
"author": {
"@type": "Person",
"name": "Noah Reed",
"url": "https://example.com/profiles/noah-reed"
},
"datePublished": "2026-03-20T16:00:00Z"
},
{
"@type": "Comment",
"text": "The manufacturer guide recommends testing the lamp with a fresh battery.",
"author": {
"@type": "Organization",
"name": "Lantern Helper Bot",
"url": "https://example.com/bots/lantern-helper"
},
"datePublished": "2026-03-20T17:00:00Z",
"digitalSourceType": "https://schema.org/AlgorithmicMediaDigitalSource"
}
]
}If your thread has 40 comments and the markup lists only 10, set commentCount to 40. Use TrainedAlgorithmicMediaDigitalSource instead for a reply written by a trained model, such as an assistant built on a large language model. These examples are displayed code. Do not paste them into a live page's own JSON-LD.
Q&A page markup before and after
The baseline Q&A page has one Question, answerCount, and one acceptedAnswer with text. In Google's feature, the question needs name, and rich result eligibility needs at least one accepted or suggested answer.
{
"@context": "https://schema.org",
"@type": "QAPage",
"@id": "https://example.com/questions/quiet-fan#page",
"mainEntity": {
"@type": "Question",
"@id": "https://example.com/questions/quiet-fan#question",
"name": "How can I make a desk fan quieter?",
"text": "My desk fan works well but makes a rattling sound. What should I inspect first?",
"answerCount": 1,
"datePublished": "2026-03-19T12:00:00Z",
"author": {
"@type": "Person",
"name": "Ari Lopez",
"url": "https://example.com/profiles/ari-lopez"
},
"acceptedAnswer": {
"@type": "Answer",
"text": "Turn off and unplug the fan, then check whether the grille, base, or screws are loose.",
"datePublished": "2026-03-19T13:15:00Z",
"author": {
"@type": "Person",
"name": "Jordan Kim",
"url": "https://example.com/profiles/jordan-kim"
},
"url": "https://example.com/questions/quiet-fan#answer-1"
}
}
}In the updated version, a follow-up question from the asker appears as a nested Comment, so commentCount is 1 and answerCount stays 1. Together they equal two replies of any type, which matches the visible page. The accepted answer in this example was written by a trained model, so it carries TrainedAlgorithmicMediaDigitalSource. If a person wrote yours, leave digitalSourceType off.
{
"@context": "https://schema.org",
"@type": "QAPage",
"@id": "https://example.com/questions/quiet-fan#page",
"mainEntity": {
"@type": "Question",
"@id": "https://example.com/questions/quiet-fan#question",
"name": "How can I make a desk fan quieter?",
"text": "My desk fan works well but makes a rattling sound. What should I inspect first?",
"answerCount": 1,
"commentCount": 1,
"datePublished": "2026-03-19T12:00:00Z",
"author": {
"@type": "Person",
"name": "Ari Lopez",
"url": "https://example.com/profiles/ari-lopez"
},
"acceptedAnswer": {
"@type": "Answer",
"text": "Turn off and unplug the fan, then check whether the grille, base, or screws are loose.",
"datePublished": "2026-03-19T13:15:00Z",
"author": {
"@type": "Organization",
"name": "Example Help Assistant",
"url": "https://example.com/assistant"
},
"url": "https://example.com/questions/quiet-fan#answer-1",
"digitalSourceType": "https://schema.org/TrainedAlgorithmicMediaDigitalSource"
},
"comment": [
{
"@type": "Comment",
"text": "Does the rattling happen at every speed?",
"datePublished": "2026-03-19T14:00:00Z",
"author": {
"@type": "Person",
"name": "Ari Lopez",
"url": "https://example.com/profiles/ari-lopez"
}
}
]
}
}The JSON-LD syntax follows the W3C JSON-LD 1.1 Recommendation, and the vocabulary comes from Schema.org, including commentCount, digitalSourceType and QAPage. Schema.org defines the vocabulary. Whether a property counts for a Google feature is set by Google's own documentation.
Update an existing implementation in one sitting
The work fits into one pass over your templates. If your platform generates the markup for you, whether that is WordPress, Shopify, Wix, Squarespace or Webflow, look at what your theme, app or plugin already outputs before adding anything by hand, so you update that output instead of adding a second description of the same thread.
- Name each template as a general discussion thread or a single-question Q&A page.
- Keep the properties Google requires for that feature.
- Add
commentCountwhere not every comment appears in the markup. - Add
digitalSourceTypeonly to content a trained model or an automated process produced. - Nest
commentobjects to match the reply relationships visitors can see. - Add
sharedContentwhere a post mainly shares a link, image, video, or quoted post or comment. - Confirm that every marked-up value appears on the visible page.
Step 7 is the one that protects you. Structured data describes visible content, so a count, a label or a reply that visitors cannot see on the page should not be in the markup.
Test the live page after deployment
Run the deployed URL through Google's Rich Results Test and read the detected items. The Schema Nova schema validator can help you review the JSON-LD itself. Keep four questions apart as you test: whether the JSON is valid, whether the vocabulary is correct, whether a specific Google feature's requirements are met, and whether the page is actually shown in search.
After deployment, use the relevant rich result report and URL Inspection in Search Console. If the report lists invalid items, fix them, inspect a live URL, request validation, and allow time for Google to crawl and reindex. For warnings that name a missing field, the guide on fixing missing field schema warnings walks through the loop.
Google says structured data can make a page eligible for a rich result, but it does not guarantee that one will appear. Correct markup is the starting condition, not a promise about display.
Make forum and Q&A markup readable beyond Google
Bing's structured data guidance describes Schema.org and JSON-LD as annotation formats, so the same JSON-LD block serves Bing, and its webmaster guidelines point to crawlable pages, visible content and accurate markup. Yandex documents questions and answers on mobile search results, with one question per page and acceptedAnswer or suggestedAnswer plus the answer text. Apple's Applebot documentation says sites can use Schema.org and Open Graph markup and offers a validator.

To reach ChatGPT search and Perplexity, allow OAI-SearchBot and PerplexityBot (see Perplexity's crawler documentation) in robots.txt, and show the full conversation in crawlable HTML. For the full comparison, read whether AI systems use schema markup.
Questions people ask
What did Google change on March 24, 2026?
Google's update log says it added more supported properties for Discussion Forum and Q&A Page markup, to give its ingestion systems more clarity on comment thread structure. The linked feature pages now cover commentCount, digitalSourceType, nested comments and an expanded sharedContent.
Are the new properties required?
No. Google lists commentCount and digitalSourceType as recommended for its feature, and the required properties stay the same. Add a property only when it accurately describes what the page shows.
Is my existing Q&A or forum JSON-LD still valid?
Existing valid markup remains usable. If it already has the required properties for its Google feature, the update does not make it invalid, and you can add the new properties template by template.
Does omitting digitalSourceType cause a problem?
Google says that when the property is omitted, it assumes the content is human-generated. Use it only on content a trained model or an automated process produced, with TrainedAlgorithmicMediaDigitalSource for a model and AlgorithmicMediaDigitalSource for a simpler process such as a reply bot.
Can one page use both DiscussionForumPosting and QAPage?
Google says not to use both types on one page just to increase coverage. Choose DiscussionForumPosting for a general discussion and QAPage for a page built around one question and its answers.
