The Schema Markup Your B2B SaaS Site Is Probably Missing

Run your home page through a schema validator and you will probably feel fine. Something comes back: WebSite, WebPage, maybe an Organization node with a name and a logo. That is your SEO plugin talking, not your company. The question is no longer whether you have schema. It is whether your schema says anything.

Schema Is Mainstream. Yours Is Generic.

In 2024, the Web Almanac found 41% of web pages carry JSON-LD structured data, up from 34% in 2022. Having schema stopped being a differentiator years ago. The distribution of types is the real story: generic wrappers dominate, and the types that describe an actual business barely register.

The type breakdown from that same 2024 crawl is the thesis of this post in one distribution. WebSite markup appears on 12.73% of mobile pages and Organization on 7.16%, while Product appears on 0.77% and Article on 0.18% (HTTP Archive, 2024)^[1]. SoftwareApplication, the type that describes what a SaaS company actually sells, does not register a meaningful share at all. Typical B2B SaaS schema markup lives at the top of that distribution, in the wrapper types, which is exactly the wrong end.

That distribution is not what SEO teams chose. It is what plugin defaults emit. WordPress SEO plugins ship a WebSite node, a breadcrumb trail, and a skeletal Organization on install, and most sites never go further. The result is a web where nearly every SaaS marketing site is marked up as “a website, by an organization.” Which is true, and says nothing.

If nobody owns the marketing site, nobody audits what the plugin generated on day one. Schema is one of the cheapest places to see that vacancy, because the gap between default and complete is visible in 30 seconds of view-source.

Does Schema Actually Matter for AI Search?

Less than the hype claims, more than zero. In 2026, Ahrefs found that adding schema to already-cited pages moved AI citations by amounts statistically indistinguishable from zero. Google states no special structured data is required for AI features. Schema’s real job is entity disambiguation and machine readability, not citation bribery.

The honest evidence first. In 2026, Ahrefs tracked 1,885 pages that added JSON-LD against 4,000 matched control pages and measured citation changes of +2.4% in Google AI Mode, +2.2% in ChatGPT, and -4.6% in AI Overviews, all within statistical noise (Ahrefs, 2026)^[2]. One caveat matters: the study only covered pages AI systems already cited heavily. Schema did not boost pages machines could already read. Whether it helps pages they cannot yet parse is still an open question, and the authors say so.

Google’s documentation points the same direction. There is no special structured data for AI Overviews, and the foundational advice includes keeping your structured data matched to the visible text on the page (Google Search Central, 2025)^[3]. Schema is not a side door into AI answers. It is part of being unambiguous about who you are and what your pages contain.

The surface it feeds keeps growing either way. In 2025, Semrush tracked AI Overviews appearing on 6.49% of keywords in January, peaking at 24.61% in July, and stabilizing near 15.69% by November (Semrush, 2025)^[4]. Roughly one query in six now resolves through a surface that reads your site as data. If you want to see what those surfaces already send you, tracking AI-referred traffic covers the referral side.

The Missing Types: Organization, SoftwareApplication, Person

The differentiating schema for a B2B SaaS site is three entities deep: a full Organization node on the home page, SoftwareApplication or Product markup on product pages, and Person entities for the real humans who author your content. All three are absent from plugin defaults. All three are cheap to add.

Start with Organization, because Google is explicit about it. In 2026, Google’s documentation states that Organization markup helps it understand administrative details and disambiguate your organization, and recommends properties most sites never fill: legalName, foundingDate, alternateName, sameAs links to your social and directory profiles, contactPoint, address, and trust identifiers like vatID (Google Search Central, 2026)^[5]. Compare that list to the name-URL-logo node your plugin emitted. In the site audits I run, that three-property node is the norm, and it is the delta between it and Google’s list that goes unwritten: the entity’s resume.

SoftwareApplication is the type category almost no SaaS company uses, which the 2024 adoption data makes visible. Marking up your product with its category, operating environment, and offer structure tells machines you are a software company, not a blog with a pricing page.

Person markup matters because entities compound. Real authors with sameAs profiles, job titles, and consistent identity across posts give machines a way to connect your content to credible humans. On a blog whose strategy depends on expertise signals, anonymous bylines are a structural waste.

FAQPage Outlived Its Rich Result

In 2023, Google restricted FAQ rich results to government and health sites and deprecated HowTo entirely. The SERP payoff died. The markup did not: FAQPage adoption roughly tripled afterward, because machine-readable question-answer structure kept its value for parsers even without the visual reward.

The history is worth knowing because it explains stale advice. Under the old regime, rich results earned real clicks: Milestone’s study of 4.5 million queries between 2019 and 2020 measured 58% CTR for rich results versus 41% without, with FAQ rich results peaking at 87% (Milestone Research, 2021)^[6]. That era ended in August 2023, when Google limited FAQ rich results to well-known authoritative government and health sites and removed HowTo from results entirely (Google Search Central, 2023)^[7].

Here is the counterintuitive part. In the 2024 Web Almanac crawl, FAQPage adoption had risen from 0.2% to 0.6% of desktop pages since the deprecation^[1]. Sites kept adding markup that earns no rich result. The plausible reason is the right one: question-answer pairs in structured form are exactly the shape answer engines consume. The decoration died; the data structure survived.

The discipline still applies. Mark up only questions the page visibly answers, in the same words. Schema that diverges from visible text is the one pattern Google’s guidelines consistently warn against.

The 30-Minute Schema Audit

Three pages tell you almost everything: your home page, one product page, one blog post. Run each through a validator, list the types you find, and compare that list to what each page actually is. The gap is your backlog, and most B2B SaaS sites can close it in a sprint.

The mechanics take half an hour. Run the three pages through Google’s Rich Results Test or the Schema.org validator. Then view source and search for application/ld+json, because validators show what parses, not what is missing. Write down every type present.

Then ask three questions. Does the home page’s Organization node carry more than name, URL, and logo? Does the product page say SoftwareApplication or Product anywhere? Does the blog post connect to a Person with a real identity? For most sites I audit, the answers are no, no, and no, and the fix is a few hours of structured writing about things the company already knows about itself.

Schema will not rescue a site machines cannot crawl or render; it is one layer of the larger structural work a marketing site needs, and that structural layer deserves its own owner. But of all the structural fixes, this is the one with the lowest cost and the clearest before-and-after. Your company has a founding date, a legal name, social profiles, real authors, and a product category. Right now, your site probably declares none of them.

Sources

1. HTTP Archive, Web Almanac 2024, Structured Data Chapter – Crawl of millions of home pages; 41% JSON-LD adoption in 2024 vs 34% in 2022; type distribution WebSite 12.73%, Organization 7.16%, Product 0.77%, Article 0.18%; FAQPage rose 0.2% to 0.6% of desktop pages post-deprecation ↩
2. Ahrefs, Does Adding Schema Increase AI Citations? (2026) – 1,885 pages adding JSON-LD Aug 2025-Mar 2026 vs 4,000 matched controls; AI Mode +2.4%, ChatGPT +2.2%, AI Overviews -4.6%, statistically indistinguishable from zero; limited to already-cited pages ↩
3. Google Search Central, AI Features and Your Website – Primary documentation; no special structured data for AI features; structured data should match visible page text ↩
4. Semrush, AI Overviews Study (2025) – 10M+ tracked keywords Jan-Nov 2025 with Datos clickstream; AI Overviews on 6.49% of keywords in Jan, 24.61% peak in July, 15.69% by Nov ↩
5. Google Search Central, Organization Structured Data – Primary documentation, updated Apr 2026; recommended properties include legalName, foundingDate, sameAs, contactPoint, address, and trust identifiers ↩
6. Milestone Research, SEO Click Curves (2021) – 4.527M queries, 2.736B impressions, 157M clicks, 175 domains, July 2019-July 2020; rich results 58% CTR vs 41%, FAQ rich results 87%; historical pre-deprecation baseline ↩
7. Google Search Central Blog, Changes to HowTo and FAQ Rich Results (2023) – FAQ rich results restricted to authoritative government and health sites; HowTo deprecated; no need to remove existing markup ↩

Written by Yasser Soliman

Technical Marketer

I've spent 5+ years embedded in marketing teams at B2B SaaS companies. I own the marketing website — performance, analytics, SEO, integrations — so your team ships without bottlenecks.