---
title: "Structured Data in WordPress: Article, FAQPage, Speakable and Person"
url: https://hostmy.blog/structured-data-wordpress/
date: 2026-09-14
modified: 2026-11-12
lang: en
author: "Aditya Sharma"
description: "What each of the four AI-era schema types is actually for, the markup for each, and the one that you should deliberately keep small."
categories:
  - "RankReady"
image: https://hostmy.blog/wp-content/uploads/2026/09/hmb-card-1863-1024x538.jpg
word_count: 1797
---

# Structured Data in WordPress: Article, FAQPage, Speakable and Person

\n*Update, 4 November 2026: we removed statistics from a third-party citation study (self-published, with sentence data from Gemini only), including three claims we could not find in it, corrected the FAQ rich result dates against Google's changelog, rewrote the Speakable section to match what Google documents, and added Pro and pricing information. The title is unchanged.*

\n\n\n
Four schema types carry most of the weight for machine readers on a WordPress blog. `Article`, `FAQPage`, `speakable` and `Person`. Each answers a different question, and adding one when you needed another is the usual reason a site has plenty of structured data that describes the wrong thing.

The questions they answer, in order: what is this page, what does it answer, which sentences suit being read aloud, and who is responsible for it. Markup is one of several separate jobs in [translating AI search advice into WordPress work](/ai-search-optimization-wordpress/), and it is the one with the clearest right answer.

Everything below is the markup plus the reason. If you have not yet checked what your SEO plugin already writes, do that first, because duplicating an `Article` node is worse than having none. The inspection script is in [adding schema without duplicating Yoast](/schema-plugin-wordpress/).

## Four types, four different jobs

| Type | Answers | Usually already written by | Keep it small |
| ---- | ------- | -------------------------- | ------------- |
| `Article` | What is this page and when did it change | Your SEO plugin | Not applicable |
| `FAQPage` | What specific questions does it answer | Nobody, unless you added it | Yes, 4 to 6 entries |
| `speakable` | Which sentences suit being read aloud (a Google beta, US and English only) | Almost nobody | Yes, 2 to 3 selectors |
| `Person` | Who wrote this and can that be verified | Partially, and usually thin | No, richer is better |

## Article tells a machine what the page is and when it changed

`Article` (or `BlogPosting`, which is a subtype and equally fine) is the base node. Your SEO plugin almost certainly writes it already, so your job is to check its fields rather than write a second one, which is also the short answer to [whether a Yoast site needs a second schema plugin](/yoast-alternative/).

`{
"@context": "https://schema.org",
"@type": "BlogPosting",
"@id": "https://yoursite.com/your-post/#article",
"headline": "Adding schema in WordPress without duplicating Yoast",
"datePublished": "2025-11-04T09:12:00+05:30",
"dateModified": "2026-08-19T16:40:00+05:30",
"author": { "@id": "https://yoursite.com/#/schema/person/author" },
"mainEntityOfPage": "https://yoursite.com/your-post/"
}`

The field that gets neglected is `dateModified`. It is the only machine readable statement your page makes about whether it is maintained.

That matters because it is the date a machine reads. A `dateModified` that has not moved in three years says the page has not been touched in three years, in a format a parser reads without ambiguity. It does not tell you what any assistant does with that date.

Check yours:

`curl -s https://yoursite.com/your-post/ | grep -o '"date\(Published\|Modified\)":"[^"]*"'`

Two identical timestamps on an old post means either nothing has been updated or your theme is not passing the modified date through. Both are worth fixing, and they have different fixes.

## FAQPage turns questions into data, and costs a heading each

`FAQPage` attaches question and answer pairs to a page as structured data rather than leaving them as prose a parser has to infer.

Google stopped showing FAQ rich results on 7 May 2026 and removed the feature's documentation on 15 June 2026 (Google Search Central changelog, read 4 November 2026). Add FAQPage markup only if the questions are on the page and useful to readers. Google's pages do not say whether other systems use it, and neither do we.

`{
"@context": "https://schema.org",
"@type": "FAQPage",
"mainEntity": [
{
"@type": "Question",
"name": "Does RankReady conflict with Yoast?",
"acceptedAnswer": {
"@type": "Answer",
"text": "No. It writes the AI-era layer and leaves the Article, breadcrumb and organisation nodes to your SEO plugin."
}
}
]
}`

Two rules on this one, and the second is the one people ignore.

First, every question in the markup must appear visibly on the page. Schema describing content that is not there is detectable and pointless.

Second, cap the count. Each FAQ entry generates a heading, so a long FAQ block reshapes the whole outline of the page. Keep heading order sequential, H2 to H3 to H4 with no skips. Volume does not help. Four to six real questions is our working range. Twenty invented ones is clutter for readers.

Answer text is worth writing to the same standard as the page body. Keep each answer's first sentence short and self contained, so it makes sense on its own. The editing method is in [writing short, single-claim sentences from the start](/write-sentences-ai-can-quote/).

## speakable marks the sentences suited to reading aloud

`speakable` is a property rather than a type, and it is the closest thing schema.org offers to pointing at a paragraph and saying "this part is the summary". Google documents it as a beta for English content and US Google Home devices, aimed at news and Google Assistant (Google Search Central, last updated 8 September 2026, read 4 November 2026). It does not promise that any assistant will read your page. More on [what each schema type is for](/schema-markup-guide-ai/).

`{
"@context": "https://schema.org",
"@type": "WebPage",
"@id": "https://yoursite.com/your-post/#webpage",
"speakable": {
"@type": "SpeakableSpecification",
"cssSelector": [".rnrd-summary", "h1"]
}
}`

The selector approach is more durable than XPath on WordPress, because themes rewrite their DOM structure more often than they rename a class you control.

Pick two or three selectors, not ten. The point is a small, deliberate set. Marking the whole article body as speakable communicates nothing, which is the same as not marking it.

The obvious candidate is a short summary block near the top of the post. Readers see it first, so it earns its place there.

## Person makes the author resolvable

Most WordPress sites emit an author as a string. A string is a name. A `Person` node with `sameAs` is an identity that can be cross referenced.

`{
"@context": "https://schema.org",
"@type": "Person",
"@id": "https://yoursite.com/#/schema/person/author",
"name": "Your Name",
"url": "https://yoursite.com/about/",
"jobTitle": "Founder, Your Company",
"sameAs": [
"https://www.linkedin.com/in/example/",
"https://github.com/example",
"https://profiles.wordpress.org/example/"
]
}`

Two details make this work rather than decorate the page.

The `@id` must be stable and must be the same value the `Article` node references in its `author` field. That is what joins the two nodes into one graph instead of leaving two unrelated blobs on the page.

The `sameAs` entries must be profiles that actually exist and actually belong to that person. Three real ones beat ten aspirational ones. Identity markup only carries weight when the prose behind it holds up, which is [how experience and authority get judged sentence by sentence](/eeat-seo-ai-answers/).

## Put it in the graph, not beside it

The single biggest implementation mistake is emitting a separate `<script type="application/ld+json">` block that repeats nodes your SEO plugin already owns.

Rank Math, Yoast, AIOSEO and SEOPress each publish an `@graph` array with `@id` references linking the nodes. New structured data should either be added into that graph, or describe something outside it.

Safe to add as a separate block: `FAQPage`, `speakable` on a `WebPage` node, `HowTo` where genuine.

Never add as a separate block: a second `Article`, a second `BreadcrumbList`, a second `Organization`. Those create contradictions, and a parser has no reliable way to pick the right one.

One URL's JSON-LD, correct and duplicated

One URL's JSON-LD, correct and duplicated

What one page should emit
Article x1
author -> Person
mainEntityOfPage -> WebPage
WebPage x1
Person x1

What two SEO plugins emit
Article x1 (plugin A)
Article x1 (plugin B)
-
author -> Person
author -> a different Person

WebPage x2

Two Article nodes for one URL is the failure mode. One canonical, one Article, one author.

The duplicate is not extra coverage. It is a second answer to a question that has one answer, emitted by a second plugin that does not know the first one ran.

The practical test is one question. Could a reader verify every claim in your JSON by reading the visible page? If the answer is no for any field, that field is a liability rather than an asset.

Verify after every change:

`curl -s https://yoursite.com/your-post/ | grep -c 'application/ld+json'`

Then run the URL through Google's Rich Results Test, which catches type errors that valid JSON hides, or reproduce the same checks offline with [a curl and Python validation loop](/check-schema-without-google/).

## The five minute version on WordPress

Hand rolling all four means filter callbacks into your SEO plugin's graph, custom fields for the FAQ pairs, a `speakable` selector that survives theme updates, and a `Person` node with a stable `@id`. All of it sits in `functions.php` until something changes underneath it.

That set is what [RankReady](https://wordpress.org/plugins/rankready-ai-llm-seo/) writes for you. It is a WordPress AI SEO plugin built to run alongside Rank Math, Yoast, AIOSEO or SEOPress, so your existing plugin keeps ownership of `Article`, breadcrumbs and organisation data while RankReady adds the AI-era layer without colliding with it. Setup takes about five minutes.

What that gives you is a page that parses cleanly and describes itself consistently. No plugin can promise that an assistant will cite, quote or show your page, and we do not.

\n\n\n\nWhere Pro fits: Article, FAQPage and Speakable schema and the author box with Person data are in the free plugin, and generating FAQs by hand is unlimited with your own AI provider key. Pro adds generation on publish, bulk generation, HowTo and ItemList schema and custom post types. Pro is a one-time payment with lifetime updates and a 14-day money-back guarantee. The current price is in the [pricing section](https://hostmy.blog/plugins/rankready/#pricing).

## The order to add them in

Doing all four at once on a site that has none is how mistakes get shipped together and become hard to unpick. Add them in this order instead.

Audit what exists. Count the JSON-LD blocks, read the types, fix any duplicate `Article` before adding anything.

- Fix `dateModified` so it reflects reality. This is one field and it costs nothing.

- Add `Person` with a stable `@id` and real `sameAs` entries. Do this once for the site rather than per post.

- Add `speakable` pointing at a short summary near the top of each post.

- Add `FAQPage` last, and only on pages where you have four to six questions somebody genuinely asked.

Steps 1 and 2 fix problems. Steps 3 to 5 add capability. Doing them in the other order means building on top of contradictions you have not found yet.

To test the markup, see [Schema Markup Validator vs Rich Results Test](/schema-validators-compared/).

## Start with the field nobody checks

Before adding any new type, run the `dateModified` check from earlier in this post across your ten best pages. It is one line, and on most WordPress sites it surfaces at least one page claiming it has never been touched since publication.

Once your schema is consistent, the layer underneath it is whether crawlers reach the page at all, which is layer one of [the whole audit, layer by layer](/five-layer-ai-seo-audit/).

Which of your four types is doing the least work right now, and would you notice if it disappeared tomorrow?