The six-level stack
The six levels that decide whether AI names you: four on your site, two in the loop you run. They fail bottom up, so fix them in order.
Six levels decide whether an engine names you. The first four are your site. The last two are the loop you run on top of it. They fail bottom up, which is why the order is not a preference.
What you'll learn
- The four foundation levels and the two loop levels
- Why the order is strict
- How a change is filed against a level, and what that tells you about your own habits
The foundation
| Level | The question | What it covers |
|---|---|---|
| 1. Access | Can AI engines reach and read your site? | Which crawlers you allow. The ones that fetch pages in order to cite them are not the ones that gather training data, and they have different names |
| 2. Structure | Can AI parse and extract your content? | Titles, headings, canonical URLs, entity markup, and weighted highest: whether the content exists without JavaScript |
| 3. Authority | Does AI have reason to trust you? | Links tying your entity to its other profiles, machine-readable contact details, review markup, freshness, a page saying who is behind the business |
| 4. Citations | Is AI recommending you today? | Real answers from real prompts, verified |
They fail bottom up. A blocked crawler makes perfect markup irrelevant. Unreadable content makes authority signals moot. This is why a readiness grade that stops at the first real failure is more useful than a long list of everything imperfect about your site.
The loop
| Level | The question | What it covers |
|---|---|---|
| 5. Change | Did you act on what the levels below showed? | A brief composed, approved by a person, published, with the assertions that will decide whether it did what it said |
| 6. Proof | Did the answers actually change? | Our own fetch confirming the change is live, then a re-measurement against the control captured when you shipped |
A level graded and left alone changes nothing. Levels 5 and 6 are the two things an audit can never do for you, and they are the reason the Changes ledger exists.
How a change is filed
Every change carries the foundation level it addresses, taken from what the analyser found rather than from the wording of its title. Where it sits in levels 5 and 6 is a matter of how far it has travelled, which is the tier it already carries.
A change nothing can classify is listed as unclassified rather than filed under a level it may have nothing to do with.
What the stack tells you about your own habits
Open the ledger and group by level. Three patterns come up repeatedly:
- Everything at level 2. You have been shipping structure fixes for months. Worth asking whether authority is now the binding constraint.
- Nothing at level 1, ever. Either access was fine on day one, which is rare, or nobody has looked.
- Plenty of level 5, almost nothing at level 6. Changes are being shipped outside the ledger, so no control gets captured and nothing can be proven. This is the most common failure and the easiest to fix.
Try this in Outercite
Ask the assistant "Is my website ready for AI answers?" for levels 1 to 4, then open the ledger and see which levels your recent changes actually addressed.
Related
Readiness
How the four foundation levels are graded, and the tier they produce.
Changes
Levels 5 and 6 in practice.
Ship and prove a change
The workflow that closes the loop.
Last updated on
How a citation is counted
What has to be true before a check becomes a counted citation, what the confidence score measures, and how to use it in a decision.
Intent classification
Every prompt is sorted into one of five intent types before anything is asked, so a buying-intent citation is never averaged with a passing mention in an explainer.