Nobody checks their schema markup unless something's gone wrong. That's just how it works. A Brentwood homeware retailer came to us convinced their drop in organic traffic was down to a Google update. It wasn't. It was a missing closing brace in a JSON-LD block that had been sitting broken for five months, quietly telling Google their product pages had no reviews at all, even though the site displayed hundreds of them to every human visitor.
The Error That Took Five Months to Notice
Here's the uncomfortable truth about schema markup: a broken implementation doesn't throw an error on the page. Customers see the site exactly as intended, star ratings and all. Google's crawler, however, reads the underlying code, and if that code has a syntax error, the structured data for that section gets silently discarded. No warning banner. No obvious symptom. Just a gradual erosion of rich results in search over weeks, which is exactly slow enough to blame on something else entirely — a slow season, a competitor's sale, anything but the actual cause sitting quietly in the page source.
In this case a developer had added a new product variant field during a site update back in April and introduced a trailing comma that most browsers tolerate but Google's structured data parser does not. Nobody tested it afterwards because, visually, absolutely nothing had changed. The products still showed star icons on the live site itself, pulled from a different, unaffected part of the theme. Only the data Google actually reads had broken.
We only found it because we routinely pull every client's live JSON-LD output through a validator before a quarterly review, rather than trusting that whatever was correct at launch has stayed correct since. Had we not done that, there's a decent chance this would still be running broken today, quietly costing clicks with absolutely nothing on the surface to suggest anything was wrong.
Why Rich Results Matter More Than People Think
Star ratings in search results aren't decorative. They're one of the few places where a listing can visually stand out on a results page dominated by ten identical blue links and a map pack. We've measured click-through rate differences of around 20 to 30 percent between near-identical listings with and without star snippets, across several retail clients. That's not a rounding error on a small site's traffic, and over a few months it adds up to a genuinely significant number of lost visits that never show up as a ranking problem because, technically, nothing about the ranking changed.
For the Brentwood retailer, losing those stars didn't change their ranking position much — they stayed around position four for their main terms. What changed was how often people actually clicked through once they got there. Rankings held. Traffic fell. Those two things are not the same, and plenty of site owners conflate them, which is exactly why this kind of problem can run undetected for so long. The owner had genuinely started wondering whether his products had simply gone out of fashion, when the real answer was sitting in a single malformed script tag the whole time.
A Quick Story About a Colchester Client With the Opposite Problem
Not every schema issue is an accident. A Colchester events company once asked us to add fake aggregate rating markup to pages with zero genuine reviews, because a competitor was doing it and "seemed to be getting away with it." We declined. Google's structured data policies are specific about this: marked-up ratings must reflect real, verifiable reviews, and manual actions for this exact violation are common enough that we see them reported in webmaster forums every month. The competitor eventually lost their rich results entirely when reported, and took months to rebuild trust with real reviews afterward. Sometimes doing it properly the slow way really is the faster route, even if it never feels that way at the time.
Where Schema Errors Usually Hide
Three places cause the vast majority of the problems we find during audits. Plugin conflicts are the most common — two SEO plugins on a WordPress site both trying to output schema for the same page, creating duplicate or contradictory markup that confuses parsers. Manual edits are the second, usually from a well-meaning team member copying code from a tutorial without adjusting it for their own site structure. Third is simple neglect: markup that was correct at launch but never updated when the page content changed, so the schema describes a product that no longer exists in that form, or quotes a price from two years ago. We'd add a fourth, less common cause: platform migrations, where a site moves from one ecommerce system to another and the schema template from the old platform gets copied across wholesale, field names and all, without anyone checking whether the new platform's templating language even outputs them correctly.
- Duplicate schema output from stacked plugins
- Hand-edited JSON-LD with syntax errors
- Stale markup describing discontinued products or old pricing
- Missing required fields that invalidate the whole block
Testing Is Not Optional, and It's Free
We run Google's Rich Results Test and Schema Markup Validator on every client site quarterly, minimum. It takes about fifteen minutes per site and it's the single highest-value, lowest-effort check in our whole process. Most business owners have never heard of either tool, which is understandable — nobody teaches this stuff unless you've been burned by it once, and most schema implementation guides online skip straight past the validation step entirely.
What surprises people is how fragile a single typo makes the whole block. One missing quotation mark and the entire schema object for that page can be ignored, not just the broken field within it. We've seen this take down product markup, review markup, and local business markup all in one go because they'd been nested into a single script tag by a developer trying to be tidy. A tidier structure, in this instance, created a single point of failure rather than removing one. We'd generally recommend separate script blocks per schema type specifically for this reason; it costs nothing in page weight and means one broken field doesn't bring the rest down with it.
Fixing It Without Breaking Something Else
The temptation once you've found a schema error is to rip it all out and start again. We'd argue against that unless the site is small. On a larger catalogue, a safer approach is validating section by section, fixing the syntax error first, then checking whether the remaining fields actually match current, genuine content on the page. A rushed fix that reintroduces fake or outdated data solves one problem and creates another, and we've seen agencies do exactly that under time pressure from a client who just wants the stars back immediately.
In the Brentwood case, the catalogue ran to several hundred products, so we worked through categories in order of traffic value, fixing the highest-earning product lines within the first week and leaving lower-priority, low-traffic listings for the following fortnight. That staged approach meant the business started seeing rich results return on its best-selling lines almost immediately, rather than waiting for a single, all-at-once fix across the entire catalogue.
If your site runs on a platform where developers come and go, it's worth having someone keep an eye on this as an ongoing job rather than a one-off audit. A proper handover process between developers should always include a structured data check, but in practice it almost never does, which is how a five-month gap like this one happens without anyone being particularly careless. That's part of what we build into our technical SEO work across Essex — not a glamorous task, but one that quietly protects click-through rate month after month. For businesses weighing up whether ongoing SEO support is worth the cost, cases like this one are usually the clearest answer we can point to.
The Bottom Line
Schema errors are invisible until they cost you something, and by the time they do, the cause usually looks nothing like the symptom. The Brentwood retailer's rankings never moved; their clicks still dropped by close to a quarter. Check your structured data properly, on a schedule, not just when something feels off, because by the time it feels off you've likely already lost months of clicks you'll never get back.
Looking for SEO agency in Essex?
Let's talk about what your business actually needs
Digital Impact Ventures provides SEO, AI visibility and digital marketing support for businesses across Colchester and Essex. Get in touch and we will talk through where your business is now and where it needs to go.