Stop shipping GEO theater.

If a crawler can't verify it, it doesn't count.

Most GEO reports hand you a soft score and a lot of confidence. That is not the game here. mygeo fetches the public surface, checks the boring root files that usually get missed, samples real pages, and leaves a paper trail your team can challenge.

Run a live check Read the method

No score theater Unknown stays unknown. The report does not invent certainty to look helpful.

Receipts included Every finding carries evidence lines so another person can check the same thing.

Runs stay linkable The output is stored as a run, not a screenshot you lose in chat two days later.

robots.txt sitemap.xml llms.txt sameAs author cues schema dates

Audit Lab

live public site inspection

stored runs

Run a live GEO check

Drop in a public URL. mygeo will fetch the page, inspect crawler-facing files, and build a run you can reopen later.

Paste a public URL to start.

Inspection path

Fetch the entry page and root files.
Parse sitemap coverage and sample live URLs.
Promote blockers before healthy checks.
Keep the raw markdown report attached to the run.

What usually breaks

The embarrassing GEO problems are rarely exotic.

They are the dull things. Root files that collapse into homepage HTML. Entity links that never made it into JSON-LD. Content pages that look complete until you ask where the date or author went.

Homepage HTML pretending to be sitemap.xml

Looks fine in a browser. Falls apart the moment a crawler requests the root file.

Robots policy quietly blocking the bots you actually care about

One platform toggle can undo the intent of your app-level `robots.txt`.

Article pages with no visible date, no author trail, no schema dates

Fine for a casual browse. Weak if you expect citations or trust signals.

Pretty reports with zero evidence under the claims

If the output cannot be argued with, it usually cannot be trusted either.

How a run reads

The report is layered on purpose.

You should not have to hunt through green checks to find the actual blockers. mygeo pushes the work items to the top, keeps the evidence ledger nearby, and leaves the raw markdown attached for sharing or review.

Layer 1

Priority queue

Failures and warnings go first, with direct actions and proof lines underneath.

Layer 2

Signal ledger

Key facts stay compact so you can read the state of the domain in a glance.

Layer 3

Raw run artifact

The markdown report is still there when you need the long form or want to share it.

Open benchmark

The board ranks transparency, not theater.

Each site row reflects its latest stored run. Ordering is explicit: fewer fails, then fewer warnings, then more passing checks. No hidden weightings. No composite GEO grade.

Benchmark board

Waiting for current benchmark data.

Open full board

Recent public runs

Waiting for recent run data.

Recurring blockers

Waiting for recurring blocker data.

Narrow checks

Use the wedge that matches the question.

Not every visitor wants a full GEO explainer. Some just want to know whether GPTBot is blocked, whether `sitemap.xml` is real, whether homepage `sameAs` exists, or whether the content pages carry authors and dates, or whether the whole crawler-facing surface holds up.

GPTBot checker Verify whether the public robots response still blocks GPTBot.

ClaudeBot checker See whether ClaudeBot is open on the live surface or only in theory.

llms.txt checker Check whether the root discovery file exists and can be fetched right now.

sitemap.xml checker Verify that sitemap.xml is real XML with URLs instead of a fallback or empty shell.

sameAs checker Check whether homepage JSON-LD links the site to real profiles or mirrors.

Content provenance checker Verify whether article-like pages expose authors, visible dates, and schema dates.

AI crawler audit Run the full deterministic audit when the question is broader than one signal.