The websites were connected. Their data needed to be too.
A founder profile, a certification service and a software product do different jobs. Across this network, though, they often refer to the same people and organisations.
That creates a practical problem. If one page gives a person a new identity, another describes a service as a company and a third repeats a claim nobody can verify, the network starts telling several versions of the same story.
Different pages gave the same person or product different identities. Some repeated the same markup. Others showed different data once the page had loaded. A few fixes also depended on a platform move or a change to the visible content.
The job was to turn that into a clear set of changes the client's team could use.
We started with the pages, not a schema template.
For each of the 25 URLs, we read the source HTML, loaded the page and checked its content, markup and available Search Console records. We checked who or what the page described, which links between brands we could support and which URL search engines should use.
The audit produced 94 issue and action records. These covered more than broken code. We also logged claims to correct, identities to check and work that needed to happen before a file could go live.
We then grouped the 17 most urgent URLs into 12 work units. That let us handle a shared template or identity once and give the client's team a joined up plan.
One identity for each person, organisation and product.
We gave the main people, brands and products a stable identifier, then set rules for pages that referred to them.
Louis Carter's profile and authored content could point to the same person. Best Practice Institute provided the organisation and publisher anchor. Most Loved Workplace's certification service and credential had distinct roles. CertCheck and VisiPage were modelled as applications, with supported connections to the rest of the network.
We checked identity links instead of copying them from existing markup. If we could not support a link or claim, we left it out of the final code.
What changed in the implementation package
Louis Carter
We brought the person, profile and website data together around one Louis Carter identity. His profile page's code named the person it described, and the article code used that same author identity.
Across the three priority pages, the original live validation baseline contained 99 errors and 80 warnings. Their corrected code files each recorded zero errors and zero warnings in Schema.org validation. The profile code also returned a valid Profile page and Breadcrumb in Google's code test.
Best Practice Institute
We prepared corrected data for both audited pages: the homepage and a Kyndryl report. The package defined one organisation and website identity. It also explained how to keep the report content and markup available after the page had loaded.
Most Loved Workplace
All five audited pages received implementation files covering the certification service, articles, company profile and ranking pages.
One ranking list contained a dummy company record with corrupted text. We removed it from the final code without inventing a replacement, then rebuilt the list as 87 entries in order with a matching total. Our notes identified the record on the page that the client's team needed to fix before that file could go live.
CertCheck
We prepared a consistent application identity and connected its homepage and sample report. The plan also covered the sitemap and internal links needed to help crawlers find those routes.
We kept the application data factual. No prices, ratings or reviews were added to fill a validator field.
VisiPage
Five priority URLs received corrected code. Four articles were aligned with the client's confirmed research URL pattern, and three subject organisations were connected to their verified official websites.
We also removed two unsupported or false FAQ claims from the final code while preserving the supported article and breadcrumb data. The package recorded the corresponding content and redirect work for the client's team.
What we delivered
A concrete implementation package for the client’s team, with a shared identity registry and checks recorded against the code.
The client received all 17 page files, a shared identity registry, rules for three reusable templates, setup notes and a tracker showing what still needed to happen.
All 17 page files passed strict JSON parsing. All 49 references between files resolved, with no conflicting entity definitions found by the package checks.
Sixteen of the 17 final code files completed Schema.org validation with zero errors and zero warnings. The remaining Most Loved Workplace ranking page could not receive an external validator result after repeated attempts, so we kept that check open.
Seven targeted Google Rich Results code tests completed with zero errors and detected 12 valid items. Six tests retained optional warnings, which stayed in the handoff record.
The client owned deployment, including the Most Loved Workplace platform cutover, redirects and visible content changes. The documented result is the audit, architecture and checked implementation package. Search performance and AI citations would need a separate comparison after deployment.