Schema markup — the structured data you embed in your site as JSON-LD — is the single highest-leverage technical change a small business website can make for AI visibility. It's also the one most likely to be missing, broken, or wrong.
Across the audits we run, the same handful of schema mistakes show up over and over again. This piece is a diagnostic list: ten errors we find constantly, why each one matters, and the specific fix.
If you're a small business owner: this is a useful checklist to hand to your web developer. If you're a developer or agency: this is the punch list to run before declaring a site AI-ready.
1. No Organization or LocalBusiness schema at all
The mistake: The site has no schema block describing the business itself. AI systems land on the homepage and have no machine-readable way to know what this is.
How often we see it: Roughly 70% of small business websites.
Why it matters: This is the foundation. Without an Organization or LocalBusiness block, every other schema choice gets discounted because the AI can't anchor it to a clear entity.
The fix: Add an Organization schema (or LocalBusiness for local service businesses, or an industry-specific subtype like Restaurant, Dentist, or HomeAndConstructionBusiness). Populate the standard fields: name, url, logo, description, address, telephone, email, foundingDate, and for local businesses, openingHours. WordPress plugins like Rank Math or Yoast SEO handle this through a setup wizard if you'd rather not hand-write JSON-LD.
2. Generic Organization schema when you should be using a specific subtype
The mistake: A dentist's site uses "@type": "Organization" when it should be "@type": "Dentist". A restaurant uses Organization instead of Restaurant. A contractor uses it instead of HomeAndConstructionBusiness.
Why it matters: Schema.org has hundreds of specific business subtypes. The more specific the type, the more confident the AI is about what kind of business you are — and the more accurately it can answer questions about you. Organization works; the specific subtype works much better.
The fix: Pick the most specific applicable type. Schema.org's LocalBusiness tree has dozens of subtypes — Bakery, Plumber, MovingCompany, BeautySalon, AccountingService, RealEstateAgent, and so on. Update the @type field. This is a one-line fix that often has outsized impact.
3. Missing sameAs links
The mistake: The Organization schema is present but has no sameAs array. AI has no way to connect your website to your Google Business Profile, LinkedIn, or social profiles.
Why it matters: sameAs is how a machine confirms "the entity on this website is the same entity as on these other sites." Without it, AI sees your homepage and your Facebook page as two separate businesses that happen to share a name. With it, they consolidate into one entity with much stronger confidence.
The fix: Add a sameAs array to your Organization schema linking to your Google Business Profile (the /g/... URL), LinkedIn company page, Facebook, Instagram, YouTube, X, Yelp listing, BBB profile, and any industry-specific directories. Five to ten high-quality links is the sweet spot.
"sameAs": [ "https://www.linkedin.com/company/your-business", "https://www.facebook.com/yourbusiness", "https://maps.google.com/...your-place-id...", "https://www.yelp.com/biz/your-business", "https://www.bbb.org/.../your-business" ]
4. No Service schema for what you actually sell
The mistake: The Organization schema is present, but there's no Service schema describing each offering. AI knows you exist but doesn't know specifically what you do.
Why it matters: AI recommends businesses for tasks. Someone asks ChatGPT "who can install heat pumps in Knoxville?" — and the AI needs structured proof that you offer heat pump installation, not just that you're an HVAC company. Each major service deserves its own Service block linked back to the parent organization.
The fix: Add a Service schema for each top-level offering — typically one per page if you have dedicated service pages. Include name, description, provider (linking back to the Organization), areaServed, and ideally offers for pricing.
5. FAQPage schema with no FAQ content (or vice versa)
The mistake: Either the page has visible FAQ content but no FAQPage schema, or the schema exists but the actual FAQs were removed from the page months ago.
Why it matters: FAQ schema is one of the most-quoted schema types in AI answers — it's literally pre-formatted Q&A that AI can pull directly into responses. But the schema and the visible content must match. Schema without matching content is a violation of Google's structured data guidelines and gets ignored. Content without schema means AI has to extract the Q&A from prose, which it does inconsistently.
The fix: Make them match. If your page shows a FAQ, add FAQPage schema with the exact same questions and answers. If you've removed FAQ content from the page, remove the schema. The Beacon audit explicitly checks for this mismatch.
6. Schema using deprecated microdata or RDFa instead of JSON-LD
The mistake: The site has structured data, but it's expressed as itemscope and itemtype attributes scattered through HTML, or RDFa attributes, rather than as JSON-LD blocks in the <head>.
Why it matters: Google has officially recommended JSON-LD as the preferred format since 2017. Most AI crawlers are tuned for it. Microdata still works, but it's harder to maintain, more error-prone, and treated as second-class by modern parsers.
The fix: Migrate to JSON-LD. If you're on WordPress, the modern plugins (Yoast, Rank Math, Schema Pro) all output JSON-LD by default — installing one and turning it on usually replaces older microdata.
7. JSON-LD with syntax errors
The mistake: The schema block exists, but it has a trailing comma, an unclosed bracket, or smart quotes copied in from Word. The whole block silently fails.
Why it matters: A broken JSON-LD block is worse than no schema — the AI sees the attempt, can't parse it, and may lower its confidence in everything else on the page. The Beacon audit checks for parse errors specifically because they're surprisingly common.
The fix: Validate before deploying. Paste your schema into validator.schema.org or Google's Rich Results Test (search.google.com/test/rich-results) and fix anything flagged red. JSON has no tolerance for trailing commas — they're the single most common error we see.
8. Schema doesn't match what the business actually is
The mistake: The schema says "@type": "Restaurant", but the business is actually a catering company. Or schema says "@type": "Dentist" on a website for a dental laboratory (the lab that builds crowns and bridges — not a clinic).
Why it matters: Mismatched schema is worse than missing schema. The AI gets a confident signal that's wrong, which corrupts its understanding of the business and leads to bad recommendations or no recommendations at all. The Beacon audit's Claude-assisted check looks specifically for this kind of misalignment.
The fix: Audit your schema against your actual business. Is the type accurate? Are the services described accurately? Does the address match a real location? If you've grown or pivoted since the schema was originally added, update it.
9. No BreadcrumbList schema
The mistake: The site has good navigation, but no BreadcrumbList schema telling crawlers how pages relate hierarchically.
Why it matters: Breadcrumb schema is the lowest-effort, highest-leverage way to communicate site structure. It tells AI "this page lives under /services/, which lives under the homepage." That hierarchical context helps AI understand the relative importance of pages and surface the right one in a given answer.
The fix: Add BreadcrumbList schema to every page below the homepage. Most WordPress SEO plugins do this automatically once breadcrumbs are enabled in their settings.
10. Review and AggregateRating schema missing — when reviews exist
The mistake: The business has Google reviews, Yelp reviews, industry-specific reviews — but none of that authority shows up in machine-readable form on the site.
Why it matters: Reviews are one of the strongest authority signals AI uses to decide whether to recommend a business. If your reviews live exclusively on third-party platforms and never appear in your own site's schema, you're leaving the signal on the table. Adding AggregateRating to your Organization or LocalBusiness schema is a small change with disproportionate impact.
The fix: Add an aggregateRating property to your Organization schema using your honest Google review stats (rating + count). Don't fake the numbers — beyond the obvious ethics, Google will catch inflation and penalize. The truth is enough if it's well-structured.
"aggregateRating": {
"@type": "AggregateRating",
"ratingValue": "4.8",
"reviewCount": "127"
}
How to find these on your own site
Three tools, in order of effort:
- Google Rich Results Test (search.google.com/test/rich-results) — paste any URL and see every schema block, plus any errors Google detected.
- Schema Markup Validator (validator.schema.org) — a more permissive validator from Schema.org itself. Catches things Google's stricter tool ignores.
- BeaconBird audit — free, and explicitly checks for all ten of these issues plus 20+ other AI-readiness signals.
Most sites have at least three of the ten. Some have all ten. Fixing them is usually a one-time engagement, not an ongoing project — which is exactly the kind of work BeaconBird does.
Common questions
Does my small business website really need schema markup?
Yes, in almost every case. Schema is the machine-readable way to tell AI systems who you are, what you do, where you operate, and what you sell. Without it, AI has to infer everything from prose — which it does poorly and often gets wrong. For local service businesses, LocalBusiness schema is close to mandatory.
How do I check what schema my site has?
Use Google's Rich Results Test (search.google.com/test/rich-results) — paste your URL in and you'll see every schema block on the page. The Schema Markup Validator at validator.schema.org is also useful and more permissive. A free BeaconBird audit checks for the most important schema types and flags what's missing.
Can I add schema myself if I'm not a developer?
On WordPress, plugins like Rank Math and Yoast handle Organization, LocalBusiness, and FAQ schema for free with a setup wizard. For anything more custom — like Service schema for each offering or Review schema — you'll either need a plugin like Schema Pro or someone comfortable pasting JSON-LD into the head section.
Will adding schema affect my Google rankings?
Indirectly, yes. Schema doesn't boost rankings by itself, but it qualifies your pages for rich results — star ratings, FAQ accordions, breadcrumb trails — which increase click-through rates significantly. Higher click-through compounds into ranking gains over time.
How long does it take for AI to see updated schema?
Once you publish, AI crawlers typically re-index within a few days to a few weeks depending on the platform and your site's crawl frequency. The qualitative effect — AI building confidence in the new entity definition — takes longer, usually weeks to a few months.