How to Use the AEO Canon as a Diagnostic
The AEO Canon is a diagnostic, not a checklist — walk its eight pillars in order and the first one you fail is your highest-leverage fix. Here is how to run the audit on your own site.
The AEO Canon is a diagnostic, not a checklist: walk its eight pillars in order and the first one you fail is your highest-leverage fix. Because the pillars form a cascade, you do not work on all of them at once — you find the highest broken gate and fix that, because nothing below it can reach an engine until it holds.
The AEO Canon · the cascade
Walk the gates in order — the first one you fail is your highest-leverage fix. Run the diagnostic →
Run the diagnostic
Step through the eight gates below, answering yes or no for your own site. The tool stops at your first broken gate and points you to the pillar deep-dive where you should start. Answer honestly rather than aspirationally — the value is in finding the real constraint, not the comfortable one.
Canon diagnostic
Gate 1 / 8
Can AI crawlers (GPTBot, ClaudeBot, PerplexityBot) fetch your content as server-rendered HTML, without being blocked or hidden behind JavaScript?
For the framework behind the questions, see the full AEO Canon and the three layers of AEO; for a one-page reference to scan alongside your result, see the Canon at a glance.
Why is the Canon a diagnostic and not a checklist?
The Canon is a diagnostic because its pillars depend on each other in a fixed order, so the useful question is not "which of these am I missing?" but "which is the first one I'm missing?" A checklist invites you to tick off the easy or interesting items in any order. A diagnostic forces you to find the one constraint that is actually capping your visibility — and to fix it before anything else.
This is the same logic as the theory of constraints in operations: a system's throughput is limited by its single tightest bottleneck, and improving anything other than the bottleneck produces no gain. AEO works the same way. Your citation rate is gated by your highest broken pillar, and polishing the pillars beneath it is motion without progress. The diagnostic exists to stop you from optimizing the wrong thing — a failure mode that is especially common in AEO, where the most visible, fun work (writing more content, adding schema markup) often sits below the actual constraint (a blocked crawler, a missing reputation).
This matters because effort spent below a broken gate is wasted. The dependencies run downward: an engine has to read you before it can choose you, and choose you before it can keep choosing you. So a break high in the stack neutralizes every strength beneath it.
The cascade, made concrete
A perfectly credible, original, freshly-updated page sitting behind a blocked crawler earns zero citations — Access failed, so none of the work above it ever reached an engine. Likewise, a fast and quotable page that answers a question nobody asks AI earns nothing — Alignment failed. The highest broken gate is always the right place to start.
What does each gate test?
Each gate reduces a pillar to a single yes/no question you can honestly answer about your site. Here is what each one is really checking, in cascade order.
- 1
Access — can a crawler read your raw HTML?
Test: fetch your page as an AI crawler would, with JavaScript off. If the content isn't there, you fail. Vercel found GPTBot executes zero JavaScript across 500M requests.
- 2
Alignment — are you answering the real question?
Test: do your headings match the full, conversational questions people actually ask AI, rather than terse keywords? If not, you're retrieved for the wrong queries.
- 3
Extractability — is your answer liftable?
Test: does each section answer its question in a complete opening sentence that makes sense out of context? If the answer is buried, the engine can't lift it.
- 4
Authority — does the web vouch for you?
Test: is your brand mentioned and corroborated on reputable third-party sites? Ahrefs found mentions (0.664) outweigh backlinks (0.218) for AI visibility.
- 5
Credibility — do you show your work?
Test: does every important claim carry an inline statistic, quotation, or named source? Evidence is the top visibility lever in the Princeton GEO study.
- 6
Originality — are you the primary source?
Test: do your best pages contain first-hand data or insight available nowhere else? If it's all reassembled generic content, the engine has no reason to cite you.
- 7
Freshness — is your content current and dated?
Test: is key content substantively updated with a visible recent date? Seer found 65% of AI crawler visits go to content under a year old.
- 8
Adaptability — do you measure and adjust?
Test: do you track citation share per engine on a fixed prompt set and treat tactics as hypotheses? Engines overlap only ~11%, so blind tactics decay.
How do you interpret your result?
You interpret the result by layer: where your first broken gate falls tells you what kind of work you need and how urgent it is. The diagnostic groups into the same three layers as the Canon, and each layer fails with a recognizable symptom.
The gate tests above trace to the same evidence as the Canon itself: Vercel's analysis of 500M crawler requests on rendering (Access), Ahrefs' study of 75,000 brands on mentions (Authority), the Princeton GEO study (arXiv 2311.09735) on evidence (Credibility), and Seer Interactive's recency research on freshness (Freshness). The diagnostic just turns those findings into questions you can answer.
A break in the Foundation layer (Access, Alignment, Extractability) shows up as broad invisibility — good content that engines simply never surface. It is the most urgent break and usually the fastest to fix, because it is technical and editorial rather than earned. A break in the Reputation layer (Authority, Credibility, Originality) shows up as being "almost cited" — retrieved and in the running, but passed over for sources the engine trusts more. A break in the Momentum layer (Freshness, Adaptability) shows up as erosion — citation share that declines over time even though the page didn't get worse. Match your symptom to the layer and the diagnosis sharpens. We unpack these layers in the three layers of AEO.
Don't ask what you could improve. Ask what you're failing first — then fix only that.
What can the diagnostic tell you — and what can't it?
The diagnostic tells you where your binding constraint is; it does not tell you that the pillars below it are perfect. This is an important distinction. When the tool stops at, say, Authority, it is saying "this is the highest gate you fail," not "everything above it is flawless and everything below it is irrelevant." You may have smaller issues in Extractability too — but they are not worth touching until Authority holds, because Authority is what is currently capping your visibility.
What the diagnostic deliberately does not do is score you. There is no "7 out of 8" to optimize, because a high score with one early break is worse than a lower score with no break high in the stack. A site that passes seven pillars but fails Access is invisible; a site that passes Access, Alignment, and Extractability but hasn't yet built Reputation is at least in the running. The number of gates you pass matters far less than where your first failure sits. That is why the tool reports a single diagnosis — your first broken gate — rather than a percentage. It is built to focus you, not to flatter you.
Be honest at each gate
The diagnostic is only as good as your answers. "Sort of" is a no. If you can't point to concrete evidence that a gate passes — a server-rendered fetch, a real third-party mention, an inline statistic — treat it as a fail. The value is in finding the real constraint, not the comfortable one.
How do you act on the diagnosis?
You act on the diagnosis by fixing the first broken gate completely, then re-running — not by spreading effort across every pillar. Once the diagnostic names your weakest link, open that pillar's deep-dive for the specific moves, make the fix, and only then descend to the next gate. This focus is the whole point: one decisive fix at the binding constraint beats ten diffuse improvements that never reach an engine.
- 1
Open the named pillar
The diagnostic links you to the deep-dive for your first broken gate. That's your starting point — not the pillar that's most fun to work on.
- 2
Make the fix and verify
Apply the pillar's specific moves, then confirm the gate now passes (e.g., re-fetch your page with JS off to verify Access).
- 3
Descend to the next gate
Re-run the diagnostic. With the top break resolved, the next constraint becomes visible — work it the same way.
- 4
Schedule a re-audit
Engines change, so re-run periodically. A passed gate can reopen when your stack or a competitor's content shifts.
Why re-run the diagnostic over time?
You re-run the diagnostic because AEO is not a state you reach but one you maintain — the eighth pillar, Adaptability, exists precisely because gates reopen. A framework change in your CMS can quietly break Access. A competitor publishing fresher research can erode your Freshness advantage on a query you used to own. An engine adjusting how it weighs sources can change which of your pages get cited.
None of these show up as an error message; they show up as slowly declining citation share. A periodic re-audit turns that invisible drift into a specific, fixable finding before it compounds. Treat the Canon the way you'd treat a health checkup — not because something is necessarily wrong, but because catching the first broken gate early is far cheaper than discovering it after your visibility has fallen. Pair the diagnostic with the Canon at a glance for a fast reference, and start the deeper work from what is AEO if the fundamentals are still new.
Frequently asked questions
- How do I use the AEO Canon as a diagnostic?
- Walk the eight pillars in their cascade order — Access, Alignment, Extractability, Authority, Credibility, Originality, Freshness, Adaptability — and stop at the first one you fail. That pillar is your highest-leverage fix, because the Canon is a cascade: effort on a lower pillar produces no citations while a higher one is broken.
- Why fix only the first broken pillar instead of everything at once?
- Because the pillars depend on each other in order. A fix lower in the stack can't compensate for a break higher up — a credible, original page behind a blocked crawler still earns nothing. Fixing the highest broken pillar first is what makes your effort actually reach an engine.
- How often should I run the Canon diagnostic?
- Periodically, not once. Because engines change frequently (the Adaptability pillar), gates you once passed can reopen — a rendering change breaks Access, a competitor's fresher content erodes your Freshness. Re-running the diagnostic on a schedule catches regressions before they cost you citations.
Last updated .