About
We work from crawl data. Not from opinion.
The technical layer under GEO is measurable. Either an agent can fetch the page or it cannot. Either the answer is in the served HTML or it is injected after load. We prefer the parts of this field that resolve to a yes or a no.
01
How the practice runs
Small, and staying that way. Every audit is run by the person who writes the report. There is no analyst producing findings for a consultant to present.
We build our own crawler rather than reselling someone else's. Off-the-shelf tools report what they were configured to check two years ago, and the agent list changes faster than that.
What we hold to
Measurement over inference
A finding is either measured or inferred, and the report labels which. Inference is still useful. It just should not arrive wearing the same confidence as a status code.
Directives are decisions
A wildcard disallow is a business decision about which corpora you appear in. It should be made by someone who knows that, not inherited from a template.
Markup must match the page
Schema that overstates what the visible text says is a liability. It is detectable, it gets discounted, and in the meantime it teaches a model something false about you.
No ranking promises
We can tell you an engine could not reach a page and now can. We cannot tell you it will cite you. Anyone who quotes a citation rate in a sales call is guessing.
The test for a finding: could an engineer reproduce it from the report alone, without asking us a question? If not, it is not finished.
02
Where we are useful
Documentation-heavy products, developer tooling, data infrastructure, and any company whose buyers ask specific technical questions before they will take a call.
We are less useful to sites with under fifty pages, where a careful afternoon by your own engineer would find most of what we would.