9 Findings From Testing 40 Cape Town Business Websites for AI Readiness

If you are evaluating a new website, compare it against the AI-built website SEO failure checklist.

Direct answer: Does schema improve website AI readiness?

In this 40-site Cape Town sample, the 20 schema-present websites achieved an average combined AI readiness score of 12.5 out of 19. The 20 schema-absent websites averaged 5.8. The difference came mainly from machine-readable identity data and stronger technical hygiene. However, only six schema-present sites used LocalBusiness schema and only two described services or offers in structured data. Schema improved machine readability, but installing basic schema did not create a complete local business entity.

AI Overview-ready summary

• Easy Rank tested 40 business homepages across Durbanville, Hout Bay, Mitchells Plain, Bellville, Woodstock and Brackenfell.
• The matched sample contained 20 websites with detectable JSON-LD schema and 20 without detectable schema.
• Schema-present sites averaged 12.5 out of 19 for combined readiness, compared with 5.8 for schema-absent sites.
• Nineteen of the 20 schema-present sites had valid JSON-LD.
• Only six of those 20 sites used LocalBusiness schema.
• Only eight provided a telephone number in schema and seven provided an address.
• Only six included geographic or service-area information in schema.
• Only two described a service, offer or offer catalogue in structured data.
• Visible location clarity remained weak in both groups: only 13 of 20 sites in each group visibly mentioned the relevant suburb on the homepage.
• Structured data supports clearer machine understanding, but it does not guarantee rankings, rich results or AI citations.

How Easy Rank conducted the study

This was a controlled follow-up to our audit of 100 Cape Town business websites. We selected 40 eligible local business homepages from the larger research pool. Twenty had detectable JSON-LD structured data and 20 had no detectable JSON-LD in their server-delivered homepage HTML.

The two groups used the same suburb distribution:

Each homepage was tested on 3 September 2026. We measured three layers:

8. Visible business identity: title, meta description, one H1, suburb, telephone, email and opening hours.
9. Structured business identity: entity type, name, URL, telephone, address, geographic coverage, social profiles, opening hours and services or offers.
10. Technical hygiene: canonical link, indexability and valid JSON-LD.

The maximum combined score was 19. This score is an Easy Rank research framework, not a Google ranking factor. It measures whether a homepage gives people and machines a clear, consistent set of business facts.

We inspected server-delivered HTML and did not execute client-side JavaScript. A website that injects schema only after rendering may therefore have been recorded as schema-absent. The study also did not attempt to prove that a website would receive an AI citation. Chatbot outputs change according to platform, location, personalisation, time and prompt wording. Measuring the source signals is more repeatable than treating one chatbot answer as a permanent result.

Source accessibility and source applicability are different checks, as our municipal election information audit illustrates.

Finding 1: Schema-present websites scored more than twice as high overall

The schema-present group averaged 12.5 out of 19. The schema-absent group averaged 5.8.

That is a 6.7-point difference in this sample. It does not prove that schema caused higher rankings or citations. It shows that the schema-present websites exposed substantially more business information in a machine-readable form and generally maintained stronger technical foundations.

Google explains that structured data provides explicit clues about page meaning. It can help Google understand information about an organisation, product, event or local business. That is different from saying the markup forces Google to rank the page or display a rich result. Our 11-point AI-agent readiness checklist extends this distinction to crawl access, visible business information and practical user journeys. The same evidence-first approach applies to search announcements: our Google UGC Fresh Data Program explanation distinguishes programme eligibility from ordinary website improvements.

Finding 2: Basic schema was common, but LocalBusiness schema was not

All 20 websites in the schema group contained JSON-LD. Nineteen had valid JSON-LD according to our parser. This highlights why correct schema markup implementation matters for local businesses.

Sixteen used some form of organisation or business entity type, but several relied on broad website and organisation markup. That can identify a publisher or organisation without describing the details a customer may need from a local business.

For a dentist in Bellville, a plumber serving Durbanville or a salon in Brackenfell, the local layer matters. Relevant markup can identify the business type, address, telephone number, opening hours and service area. Google’s LocalBusiness documentation specifically supports business details such as hours and departments.

Finding 3: Contact details were visible more often than they were structured

In the schema-present group, 17 of 20 websites displayed a detectable telephone number, but only eight included a telephone number in structured data.

The gap matters because visible and structured information serve related but different purposes. A customer can read a number in a header or footer. A crawler benefits when the number is also attached to the correct business entity through a property such as telephone.

The schema-absent group performed worse for visible telephone clarity, with 12 of 20 showing a detectable number. This suggests that weak structured implementation sometimes sits alongside weak visible business information.

Finding 4: Address and service-area data were missing from most schema

Only seven of the 20 schema-present sites included an address in their structured data. Only six included geographic coordinates, an area served or another service-area field.

This is a serious local clarity gap. Strong local SEO foundations help businesses appear more relevant in these searches. “Electrician Cape Town” and “electrician Brackenfell” can reflect different urgency, proximity and travel expectations. A service business that works across Durbanville, Bellville and nearby northern suburbs should explain that coverage visibly and consistently.

The same principle applies to Hout Bay, Woodstock and Mitchells Plain. A vague “Cape Town” reference may be technically true, but it gives customers less confidence about whether the business actually serves their area.

Finding 5: Service information was the weakest structured layer

Only two of the 20 schema-present websites used detected properties for a service, offer or offer catalogue.

This was the most commercially important gap in the study. A business website does not exist only to state a company name. It must explain what the business does, for whom and where.

A generic Organisation or WebSite graph may establish identity, but it does not necessarily define individual services. A builder may offer renovations, extensions and new builds. An accountant may provide tax returns, payroll and company compliance. A dentist may provide examinations, X-rays and restorative care. Those service relationships should be clear in visible content first, then represented accurately in schema where appropriate.

Finding 6: Social identity connections were better, but still incomplete

Thirteen of the schema-present sites included sameAs or an equivalent social identity connection. This was one of the stronger structured fields.

Correct social profile links can help disambiguate an organisation, particularly when a business has a common name. They should point to authoritative profiles controlled by the same organisation. Random directories and unrelated profiles should not be added merely to make the schema appear larger.

Entity consistency matters more than volume. The name, URL, logo, telephone number and location should agree across the website, Google Business Profile and genuine social profiles.

Finding 7: Location wording was equally weak in both groups

Only 13 of 20 websites in each group visibly mentioned the suburb used to identify the business for this study.

This result is important because schema cannot repair vague visible content on its own. Google’s structured-data guidelines require markup to represent the main visible content accurately. If a website claims a wide service area only inside JSON-LD while the page never explains that coverage to visitors, the implementation becomes harder to trust and maintain.

Strong local pages should naturally explain where the business is based, which suburbs it serves and whether customers visit a physical location or receive a mobile service. The wording should be useful, not a block of repeated suburb names.

Finding 8: Canonical implementation separated the groups

All 20 schema-present websites had a detectable canonical link. Only 12 of 20 schema-absent websites had one.

Businesses should also regularly track AI search visibility measurement signals when similar or duplicate versions exist. It does not guarantee indexing, but its absence can create avoidable ambiguity, especially when a site is available with tracking parameters, alternate paths or inconsistent URL versions.

Both groups appeared indexable in the server-delivered homepage HTML. None of the 40 homepages contained a detected noindex directive during the audit.

Finding 9: Visible heading structure remained poor in both groups

Only nine schema-present websites and five schema-absent websites had exactly one detectable H1.

This shows why schema should never become a substitute for on-page optimisation. A clear H1 helps visitors and search systems understand the page’s main topic. Multiple H1 elements are not automatically a penalty, but missing, duplicated or generic headings often signal that the page hierarchy was not planned carefully.

Meta descriptions performed better. Seventeen schema-present websites and 15 schema-absent websites had a detectable meta description. The small difference suggests that basic metadata has become common while deeper entity and service modelling remains neglected.

What these findings mean for Cape Town SMEs

The study reveals a two-layer problem.

First, many websites still fail to state basic local business information clearly. A visitor should not need to search several pages to confirm whether a company serves Mitchells Plain or Hout Bay. The homepage should establish the business name, primary service, location, service area and next action quickly.

Second, websites that do use schema often stop too early. A plugin may automatically create WebSite and Organisation markup. That is useful, but it does not mean the business has a complete local entity graph. The business may still lack its telephone number, address, opening hours, service area and service relationships.

This is where cheap websites and rapid AI-built websites can create hidden costs. The problem is not the tool. WordPress, custom code and AI website builders can all produce strong or weak outcomes. Alongside public identity checks, use the website handover checklist to verify the operational access and recovery arrangements behind the site. The risk appears when launch speed and visual polish replace content planning, technical validation and local search optimisation. A rebuild should preserve useful business information as well as improve structure. Our redesign SEO planning guide supplies a URL map and release checks.

A R3,000 website that needs structural repairs, new service content, analytics, schema and local optimisation may become more expensive than a properly scoped build. The owner also loses time while customers and search systems encounter an incomplete representation of the business.

A practical AI readiness checklist for local businesses

1. Confirm the visible business identity

Check that the homepage clearly states the business name, primary service and location. Use language a customer would understand. Avoid vague claims such as “quality solutions for every need”.

2. Define the service area accurately

State whether customers visit your premises or whether you travel to them. Mention relevant suburbs naturally. A Durbanville-based provider might also serve Bellville and Brackenfell, but those claims must reflect real operations.

3. Build a consistent entity

Align the name, address, telephone number, website URL and logo across the website, Google Business Profile and controlled social profiles. Fix conflicting details before adding more markup.

4. Use the most appropriate schema type

Choose a valid, specific business type where one exists. Do not select a type because it sounds more impressive. The schema must describe the visible business truthfully.

5. Connect services to the business

Create useful service pages with clear descriptions, audience needs, locations, proof and contact paths. Add suitable Service or offer relationships only when they match the page content.

6. Validate technical foundations

Check the canonical URL, indexability, XML sitemap, internal links, mobile layout, page speed and structured data. Use Google’s Rich Results Test and URL Inspection where relevant.

7. Measure rather than promise

Track organic queries, calls, forms, direction requests and landing-page engagement. Schema can improve understanding and eligibility, but Google explicitly states that correct structured data does not guarantee a rich result.