RomeoSEO

RomeoSEO11 min read

Google's Rich Results Test: Fix What It Finds

How to run Google's Rich Results Test: URL vs code mode, errors vs warnings, fixing schema errors, and monitoring rich results in Search Console.

Cover image for How to Use Google's Rich Results Testing Tool (and Fix the Errors It Finds)

Summary of the Article

Google's Rich Results Test tells you whether a page's structured data is eligible for rich results, and what's blocking it if it isn't. Use URL mode for anything that's already live, code mode for drafts, then fix the errors it reports and re-test.

What the Rich Results Test actually checks

The Rich Results Test is Google's official tool for checking whether a page's structured data qualifies for rich results: the extras you sometimes see in search listings, like star ratings, recipe steps, event dates or breadcrumb trails.

Two things happen under the hood, and knowing them explains most confusing results:

  1. The tool fetches the page and renders it, so it sees the final HTML your visitors see, not just the raw source.
  2. It reads the structured data on that rendered page. It supports all three formats Google accepts: JSON-LD (a script block, the most common), Microdata (attributes woven into HTML tags) and RDFa (similar, attribute-based).

The report shows which rich result types it found, the issues in each, and a code explorer that jumps to the exact spot in the rendered source when you click an issue.

Step 1: Pick the right mode (URL or code)

The landing page offers two inputs. Choose by where your page lives right now.

Use URL mode for anything live. Paste the full URL. The tool crawls the page as Google-InspectionTool, Google's own crawler identity, and tests the rendered result.

Use code mode for drafts. Paste your JSON-LD or HTML and click Run test. You can edit and re-run as often as you like, which makes this the fastest loop while writing markup.

Code mode has one big blind spot: your pasted code is tested on its own, without the rest of your live page. If a plugin or script on the live page injects your schema at runtime, code mode can't see it. Test the live URL for anything generated by JavaScript.

Pick the user agent deliberately. Below the input box you can choose smartphone (the default) or desktop. Google uses mobile-first indexing, so the smartphone default is right for almost everyone. Switch to desktop only when you suspect the desktop rendering behaves differently.

Staging and localhost pages need a public door. The tool can't reach pages behind a firewall, password protection, or your own machine. Google's own guidance: either expose a tunnel (a public URL that forwards to your local or staging server) or test the code snippet directly. Every page resource must also be reachable by an anonymous visitor; anything login-gated silently fails.

Step 2: Read the top-line status

When the test finishes, the status line sorts into a handful of values. The Google help page documents the full set:

Step 2: Read the top-line status: N valid items detected; N valid items with warnings detected; N items detected; No items detected; URL cannot be crawled; Structured data with syntax errors detected

Step 3: Tell errors from warnings (and from silence)

Open each detected item and you'll see two kinds of issues, and they're not equal:

Here's the trap most guides skip: some missing properties produce no warning at all. Article markup without datePublished or dateModified often comes back clean, even though the missing dates weaken how Google can use the page. A green result means "no detected errors for eligibility," not "nothing worth improving." Treat the test as a floor, not a ceiling.

Step 4: Fix crawl failures first

If the status is "URL cannot be crawled," nothing about your schema matters yet. The tool visits as Google-InspectionTool, not with your credentials, so robots.txt rules apply. The common access errors, straight from Google's documentation:

Step 4: Fix crawl failures first: Blocked by robots.txt; noindex; Invalid SSL certificate; DNS server unresponsive / DNS error; Hostload exceeded; Server connection error / invalid response

If robots.txt blocks a resource the page needs, you'll get a page-loading warning instead. A blocked image or CSS file can change how Google renders and understands the page, so unblock resources that matter to rendering.

One quirk to expect: with unloadable resources or rendering hiccups, two runs of the same URL can differ slightly, because the set of resources loaded varies per run. If results change with no change from you, look for a "page loading issues" warning in the report.

Step 5: Fix syntax errors in items of unknown type

When a syntax error stops the parser before it can even tell what type you meant, you get a section called "Syntax errors in items of unknown type." The usual suspects:

Step 5: Fix syntax errors in items of unknown type: Invalid JSON document; Parsing error; Incorrect value type; Duplicate unique property; Bad or empty escape sequence

A JSON formatter makes these obvious in seconds. Fix the first reported error and re-run: one mistake often produces several messages, and clearing a parse error can reveal real errors that were hidden behind it.

One subtlety worth knowing: the test tool ignores comments inside JSON-LD blocks, but that tolerance isn't part of the JSON-LD standard, so the same commented code can fail elsewhere. Google's advice is to remove all comments before publishing.

Step 6: Fix the common per-feature errors

Once the JSON parses, the errors become type-specific. The recurring patterns:

Fix one error, re-run the test, repeat. That loop beats a big-bang rewrite, because you can see exactly which change moved the result.

Step 7: Validate beyond Google's test

Google's tool checks eligibility for its rich result types. Before deploying, it's worth a second pass that checks the JSON-LD itself against the required properties, especially when a CMS or plugin is writing the markup for you.

Our free schema markup validator validates JSON-LD on any URL or pasted snippet, which is handy when you want a second opinion or need to check markup types Google's tool doesn't cover. And if the test found nothing because your markup is missing or half-built, our free schema markup generator produces valid JSON-LD for common page types to start from.

Step 8: Monitor at scale in Search Console

The Rich Results Test is one page at a time. Search Console's rich result reports under Enhancements show the same issues across your whole site, one report per rich result type, once Google finds valid markup of that type.

Three things to know when reading those reports:

The full workflow loop

Put together, the working routine looks like this:

  1. Test the live URL (or the draft snippet) with the Rich Results Test.
  2. Fix errors first, warnings second, using the code explorer to find each issue.
  3. Redeploy and re-run the test on the live URL.
  4. Watch Search Console's enhancement reports for site-wide confirmation, and use Validate fix where available.
  5. Re-run the test after every theme, plugin or template change. That's when markup breaks with nobody noticing.

Then go one layer wider. Once eligibility is sorted, the free on-page SEO checker grades a page for a target keyword, covering indexing and on-page signals beyond structured data. And if you're working through a broader ranking recovery, this test slots naturally into an SEO content audit; monitoring what happens after the fix pairs well with our guide to tracking keyword rankings in Search Console.

Common questions

The test says my page is valid. Why don't I see a rich result? Valid means eligible, not guaranteed. Google decides when and whether to show rich results, based on the query, the device and its own quality signals. A clean test is a prerequisite, not a promise.

Can I test password-protected or AMP pages? Not directly. The tool needs everything reachable by an anonymous visitor, so either expose a tunnel for staging or test the code snippet. AMP pages work through URL mode only if they're publicly accessible; some rich result types also have AMP-specific requirements, so check the feature's documentation.

How long until my fix shows up? The test updates the moment you redeploy. Search Console's reports wait for Google's next crawl, which can take days to weeks depending on the page. Validate fix speeds up re-checking of reported items.

The test is clean but traffic didn't change. What now? Rich results affect how a listing looks, not whether the page ranks. Look at impressions and clicks in Search Console for the affected queries before assuming the schema was the lever.

The next step

Run the test on your five most valuable pages today. Anything with an error goes on the fix list; anything with only warnings gets the recommended properties added when the data is real. That single afternoon usually clears more eligibility blockers than a month of guessing.

If you want to move from checking pages to fixing them, RomeoSEO finds the pages losing ground from your Search Console data and a crawl of your site, including missing or broken structured data, and writes the fixes. Start your free trial and see what it surfaces on your site.

Give Romeo a website to judge.