Discover guide

SchemaX Discover is the engine that turns your website into a machine-readable business profile: the structured layer that lets ChatGPT, Perplexity, Google AI Overviews, and other AI systems understand and recommend you.

How it works

Discover scans your site’s pages and extracts the facts that matter (business name, services, locations, contact details, evidence of expertise). Facts it can ground in your live pages publish automatically; anything uncertain or sensitive is held for you in the review queue. You spend attention only on those exceptions, then generate a profile release and deploy.

The key difference from traditional schema markup: Discover is ongoing. Your site changes, and your profile should change with it.

The review queue

After each scan, the Review queue shows every page SchemaX processed, grouped by confidence:

Auto-published: SchemaX grounded these fields on your live page and published them automatically
Needs attention: one or more fields are uncertain or conflicting, so they are held for you
Held back: the page was unclear, uncrawlable, or not worth profiling

Work the queue from the top. The confident, grounded facts are already live; your attention goes to the pages flagged for attention, starting with your homepage and key service pages, since these carry the most weight for AI recommendation.

Correcting and holding back

When you review a flagged page:

Confirm: the grounded value is accurate. SchemaX keeps it in the next release.
Override: you know better. Type the correct value and confirm. SchemaX learns the correction.
Hold back: the page genuinely doesn’t belong in the profile. Remove it from the release.

Common things worth overriding:

Business name inconsistencies (abbreviated on one page, full name on another)
Service descriptions pulled from navigation labels instead of the actual service page
Phone numbers or addresses that are outdated

Generating and deploying a release

When the queue is clean (or clean enough for a first release), go to Releases and generate a new profile release. Each release is versioned, so you can roll back to any previous release if something goes wrong.

Deploy via the managed injector, GTM, or a manual script. After deploying, run Verification to confirm the live site is serving what you expect.

Staying healthy over time

Discover works best as a routine, not a one-time setup:

Schedule scans after meaningful content changes (new service pages, location changes, pricing updates)
Respond to drift warnings quickly; drift means your live profile no longer matches what’s on the site
Re-review held-back pages periodically; sometimes a page improves and becomes worth profiling

When the scan isn’t working well

If SchemaX keeps returning weak or held-back suggestions for the same pages, the problem is usually the page itself, not the scan:

The page may not be crawlable (check robots.txt and JavaScript rendering)
The page content may be too thin or too generic to extract reliable facts
Service names or business facts may be inconsistent across the site

Fix the source first. Discover works best when there’s a clear, consistent truth on the page to extract from.