Schema markup can help your business site or hurt it, depending on how it gets done. Here’s what it is, when it’s worth investing in, and when to skip it entirely.
Most small business websites have no schema markup. Most of the ones that do have schema markup that makes things worse.
This is where many founders’ marketing strategy runs into a wall they can’t see. Every SEO article tells you to add schema. Every plugin promises to handle it for you. And yet the biggest empirical study published this year found that pages with generic, plugin-default schema were cited by AI search systems less often than pages with no schema at all.
If you run your own website on WordPress, Squarespace, Wix, or Shopify, and you’ve been told repeatedly to add schema without ever quite understanding what it is, this post is for you. I’ll cover what schema actually does, when it’s worth your time, when to leave it alone, and why the honest answer for most founder-led businesses isn’t what you’ve been told.
What Schema Markup Actually Is
Think of the nutrition facts label on a cereal box. The front of the box is for humans. You read the name, the picture, the marketing copy. The label on the side is for machines and regulators. Same information, repackaged into a standardized format that any system can parse line by line. Calories in one field. Sodium in another. Serving size in its own spot.
Schema markup is the nutrition facts label for your website. It’s a block of code that sits alongside your visible content and tells search engines, AI systems, and other machines exactly what the content means. Not just “Dr. Smith works here” but “Dr. Smith is a Person who holds the jobTitle of Dentist at a LocalBusiness of type DentalClinic at this PostalAddress with these openingHours.”
The vocabulary was created in 2011 by Google, Microsoft, Yahoo, and Yandex working together. It’s called Schema.org, and it contains about 800 types and 1,500 properties. The format Google recommends for using it is called JSON-LD, which stands for JavaScript Object Notation for Linked Data. You don’t need to know what that means. You only need to know that when you see a block of code starting with <script type=”application/ld+json”>, that’s schema markup, and Google prefers it over the alternatives because it’s self-contained and easy to maintain.
What schema does not do is also worth naming directly. It does not rank your site higher. Google’s John Mueller has said this on record, repeatedly, since at least 2019 and as recently as early 2026. Google has put the same message in plainer language through its Search Liaison communications. Using schema doesn’t give you a ranking boost. It can help you be eligible for certain displays or enhancements, but it doesn’t boost you to the top of results.
That’s the first thing to get clear. Schema is not a ranking factor. What it does is make your content eligible to appear differently. Star ratings. Price displays. Rich snippets in knowledge panels. Citations in AI answers. Being eligible is not the same as appearing, and appearing is not the same as ranking.
Why Schema Matters More in 2026 Than It Ever Has
For most of the 2010s, schema was an optional upgrade. You could make a case for it or against it, and plenty of well-performing sites ignored it entirely. That stopped being true somewhere in 2024.
The business context has shifted. 94% of B2B buyers now use large language models during the buying process, according to 6sense’s 2025 Buyer Experience Report. The winning vendor comes from the buyer’s Day-One shortlist 95% of the time, up from 85% the year before. If you’re not on the initial shortlist, the deal is almost certainly lost before you ever hear about it.
Google’s AI Overviews have changed the mechanics of search itself. Seer Interactive’s analysis, published November 2025, found organic click-through rate fell from 1.76% to 0.61% on informational queries where an AI Overview appears, a 61% decline. More than half of all Google searches in the United States now end without the user clicking on anything. The content that does get surfaced is the content that AI systems can parse, verify, and cite with confidence.
This is where schema’s role has changed. It’s no longer a nice-to-have that unlocks rich results in traditional search. It’s become the machine-readable layer that decides whether your brand exists as a distinct, recognizable entity in Google’s Knowledge Graph, or as an ambiguous string of text that could mean anything.
Microsoft has confirmed the shift directly. In March 2025, Fabrice Canel, Principal Product Manager for Bing, told an audience at SMX Munich that schema markup is used by Microsoft’s LLMs to understand content for Copilot. Google has said similar things on record. At Google’s Search Central Live event in Madrid in April 2025, John Mueller spoke about structured data’s continued importance in an AI search world and recommended focusing on schema that describes elements actually visible in search results. Google’s official 2025 guidance on succeeding in AI search includes structured data as one of the foundational recommendations.
When schema is implemented correctly, real case data still supports the older argument. Google’s own case studies cite Rotten Tomatoes showing a 25% higher click-through rate across 100,000 structured pages, Nestlé seeing 82% higher CTR on rich-result pages versus non-rich ones, and Food Network reporting a 35% visit increase after enabling search features on 80% of pages. These are real numbers from real sites, published by Google, not agency claims recycled through vendor blogs.
The operative word is correctly. This is where most founder-led businesses come unstuck.
In February 2026, an independent study by Kurt Fischman at Growth Marshal analyzed 1,006 pages across 75 commercial queries and tracked 730 AI citations across ChatGPT and Gemini. The findings, published on Zenodo with full methodology disclosed, were uncomfortable for the SEO industry. Pages with generic, plugin-default schema (Article, Organization, BreadcrumbList with minimal attributes) landed at a 41.6% AI citation rate. Pages with no schema at all hit 59.8%. Pages with attribute-rich schema, populated prices, real ratings, detailed specs, and entity-linking via Wikidata and sameAs reached 61.7%.
The takeaway is not that schema doesn’t work. It’s that thin, incomplete, auto-generated schema is a signal of low-effort content to systems that are now sophisticated enough to read the difference. Turning on Yoast’s or Rank Math’s or Squarespace’s default schema output is not a neutral action. It’s a weak signal broadcast to an increasingly skeptical audience of machines.
Fewer than 4% of schema-present pages in the study used genuine entity-graph schema (cross-page @id references, Wikidata sameAs links, nested entities with proper relationships). The remaining 96% were doing the equivalent of submitting a nutrition label that reads “Contains food. Calories some.”
For a DIY founder, this finding has an immediate practical consequence. The question is not should I add schema. It’s can I add schema that’s actually good, or am I going to end up in the generic 96%.
When Schema Is Worth Your Time and When to Skip It
Schema investment tends to pay off when three or more of these conditions apply to your business.
The first is physical presence. If you have a location or a defined service area, and your business could show up in Google Maps or the local pack, LocalBusiness and Organization schema feed the whole local ecosystem. This is still one of the highest-return applications of structured data for SMBs.
The second is category risk. If you work in a YMYL category (health, legal, finance, home services, anything where trust and credibility matter heavily for the buyer), Organization, Person, and Service schema directly inform how AI systems assess whether to cite you on high-stakes queries.
The third is authored expertise. If you publish content under the name of a real person with identifiable expertise, Article or BlogPosting schema with nested Person schema, linked to LinkedIn, credentials, and a Wikidata entry if you have one, is one of the strongest levers for AI citation of thought leadership.
The fourth is real catalog data. If you have a product or service catalog with real attribute data (populated prices, inventory, specs, and genuine third-party reviews), this is the one category where the Growth Marshal study found measurable AI citation lift from properly implemented schema.
The fifth is brand disambiguation. If your brand name is ambiguous or easily confused with something else, Organization schema with a well-populated sameAs array (Wikipedia, Wikidata, LinkedIn, Crunchbase, verified social profiles) is the primary technical path to a Knowledge Panel.
The sixth is review volume. If you have at least 10 legitimate third-party reviews from real customers, AggregateRating markup can earn star displays. Below that threshold, it looks thin and often gets suppressed. Since 2019, Google has disallowed self-reviewed star ratings for LocalBusiness and Organization, so the reviews must come from somewhere other than your own site.
The seventh is search position. If you already rank in the top five for your primary commercial queries, schema amplifies what’s working. Growth Marshal’s regression found rank position is the dominant predictor of AI citation, with an odds ratio of 0.762 per position. Moving from position five to position two produces more AI citation lift than any schema intervention tested. If you’re not ranking, schema is downstream work.
The inverse is just as important, and most SEO content won’t tell you this part. Here are the conditions that mean you should put schema off, or skip it altogether.
Pre-revenue or pre-fit businesses should wait. Your offers, pricing, and positioning are going to change. Anything you mark up today will be wrong in six weeks. Spend the time talking to customers instead. Businesses with a rebuild, migration, or major pivot planned in the next 6 to 12 months should wait. Migrations break URL structures, often change CMS, and force re-validation. Wait until the new site has been stable for 90 days after the dust settles.
Businesses with no real content library should wait. Schema is the label, not the product inside. Without real content, reviews, case studies, and testimonials to mark up, you can only produce generic markup, which is the 41.6% category we just talked about. Build the content first. Businesses with content quality issues should fix those first. Schema amplifies whatever signal it describes. If your content strategy is producing thin or low-quality material, schema only makes the problem more legible to the systems judging you.
Businesses with nobody accountable for maintenance should skip. Prices change. Availability changes. Reviews accumulate. Schema deprecates. Without a named owner auditing quarterly, schema rots into a content-parity violation, which is worse than never having it. And any schema type that’s deprecated should be skipped. This matters more in 2026 than it ever has.
Google has been aggressive about pruning schema types that no longer produce value. FAQPage rich results were restricted to authoritative government and health sites in August 2023. HowTo rich results were fully deprecated in September 2023. Sitelinks Search Box was removed globally in November 2024. In June 2025, Google deprecated seven additional types in a single announcement (Book Actions, Course Info, Claim Review, Estimated Salary, Learning Video, Special Announcement, and Vehicle Listing). Practice Problem was deprecated in November 2025. Search Console support ended for Practice Problem, nutrition facts, nearby offers, and TV season selector in January 2026.
If you spent hours building FAQ schema on your business’s site last year expecting rich results, you were working for nothing the whole time. If you’re using a plugin that still emits HowTo schema because it was built before 2023, you’re announcing to Google that your site is running outdated tooling. This is the cost of not having someone accountable for maintenance.
What DIY Founders Get Wrong
Every CMS on the market will auto-generate some amount of schema for you. None will auto-generate schema that’s actually good. The honest picture on the major ones matters here because most founders assume “my platform handles this” and stop thinking about it.
WordPress with Yoast produces a connected graph across WebSite, WebPage, Organization, and Article types, which is better than what most plugins do, but it stops at the surface. BlogPosting isn’t auto-detected, Organization is skeletal, and sameAs is usually blank unless you fill it in manually. WordPress with Rank Math offers a wider menu of types, about 20 or more, but outputs flat template-driven schema with minimal entity linking, no native cross-page @id graph, and a documented risk of duplication when stacked with a theme or another SEO plugin.
Squarespace auto-injects Organization, WebSite, BlogPosting, Event, and Product schema by default. You cannot edit or remove what it outputs. Anything beyond the defaults has to go into a code injection block in the site header, which requires hand-written JSON-LD. Wix ships basic structured data for blog, product, and event content with a limited markup panel. There’s no custom schema builder. Advanced entity work requires third-party embeds.
Shopify’s Product schema depends on your theme. Dawn and modern themes emit reasonable output, but collection pages typically get only Organization with no ItemList or BreadcrumbList. Review and AggregateRating usually requires an app, which is the single most common source of duplicate and conflicting Product schema on Shopify stores. Webflow does not auto-generate any JSON-LD. You do all of it yourself, or none at all.
The eight most common mistakes I see when founders implement schema on their own sites follow a pattern. The first is schema that doesn’t match the visible content on the page, which is Google’s explicit trigger for a structured data manual action. The most frequent version is AggregateRating injected on pages that don’t actually display any reviews. The second is duplicate schema coming from a theme, a plugin, and an app at the same time, where conflicting values can invalidate all three.
The third is stale schema that was correct when you published it and has been wrong ever since your prices, hours, or availability changed. The fourth is deprecated types still being emitted by outdated plugins. FAQPage is the classic example, still injected by dozens of Shopify apps and WordPress plugins onto sites that haven’t qualified for its rich results since 2023. The fifth is missing required properties, like Product schema without priceCurrency or hasMerchantReturnPolicy, Event schema without location, Article schema without author. Any of these make the markup ineligible for rich results.
The sixth is using the wrong type for the actual entity (generic LocalBusiness instead of Dentist, Restaurant, or LegalService, or Article markup on product pages). The seventh is missing sameAs and orphan Organization schema. Fewer than 4% of sites implement the entity-linking that actually moves the needle. This is usually the highest-leverage property people skip. The eighth is marking up review content that didn’t come from real third-party users. This is another explicit Google manual-action trigger, and it can strip rich-result eligibility across your entire site until you file a reconsideration request and Google approves it.
If you’re not sure which of these apply to your site, I built a free Schema Analyzer that will scan what your CMS is actually outputting and flag the issues. It takes about thirty seconds and requires no technical knowledge. You’ll know whether your site is producing the generic 41.6% version or something closer to the attribute-rich 61.7% version before you decide whether to fix it yourself or bring it to an audit.
The penalty structure matters. A structured data manual action does not lower your rankings. It strips all rich-result eligibility across your site until the problem is fixed and Google approves a reconsideration request. Rich results often don’t return quickly after approval. It can take months to recover.
After Google’s deprecations, seven schema types still carry their weight for most founder-led businesses. Organization schema establishes your brand as a disambiguated entity in Google’s Knowledge Graph. LocalBusiness schema (and its specific subtypes like Dentist, Restaurant, and LegalService) drives visibility in the local pack and Google Maps. Article or BlogPosting schema with nested Person markup supports AI Overview citations and Top Stories eligibility. Product schema retains full rich result support. Review and AggregateRating schema, sourced from legitimate third-party reviews, produces star-rating rich results. Service schema with a defined hasOfferCatalog gives AI systems explicit offering signals. BreadcrumbList is universally supported and benefits from sitewide consistency. Person schema for authors, founders, and practitioners links content to verifiable expertise through sameAs.
Everything else on Schema.org’s list of 800 types is either niche, deprecated, or unsupported for Search in 2026. If someone is selling you complete schema implementation covering all 50 types, they’re selling you an invoice, not value.
What to Do Next
Schema is real, necessary infrastructure in 2026. It’s also easier to do badly than well. If you can audit your current state, identify the right types for your specific business, populate them with real attribute data, cross-link them with @id references, anchor your brand to Wikidata, maintain them as content changes, and catch the deprecations before they become penalties, do it yourself.
If you can’t, this is exactly the kind of thing a proper audit reveals. Most of the businesses I work with come in thinking their marketing problem is a volume problem (more content, more leads, more channels). A surprising percentage of the time, the real problem sits below what they can see. If you’d like to know which of those things applies to your business, book a Marketing Audit. It’s $2,500, delivered in ten business days, and it covers schema alongside positioning, offer, content, technical discoverability, and marketing analytics, which are the real layers where most marketing strategy problems actually live.