The Citability Audit: Scoring a Live Page Against the 5 Principles

Marketing Academy · Field Work●Copywriting

All projects·From: Writing for AI Search: Copy That Gets Cited by ChatGPT and Perplexity

MiniAudit· 25 minutes

Adyen

Objective: Given a live blog post, score it against the lesson's 5 principles of AI-citable copy and flag which of the 3 common mistakes are present.

You're auditing an Adyen blog post about payment authorization rates before a content refresh, using the lesson's framework instead of guessing why it never appears in AI Overviews.

Score the page 0-5 against the 5 principles, one point each, and separately flag any of the lesson's 3 named mistakes present in the copy.

Before you start

What you'll need

Free path (everything below is enough to finish)

Google Sheets

FreeScore the page against the 5 principles and log flagged mistakes

Free, easy to hand to a content team as a prioritized fix list

The process

2 steps

Step 01 of 02

Using clear structural formatting AI models can parse

Headers, bullet lists, numbered steps, and tables are machine-readable signals to AI models about what information matters; a 500-word paragraph is less likely to be cited than the same data structured as a list.

The Adyen post has one 480-word paragraph explaining factors that affect authorization rates. Does this pass Principle 4?

Google Sheets— Build a 5-row scoring tracker, one row per principle, with a 0/1 score and a note.

Procedure

Read the section covering authorization-rate factors
Check whether the factors are presented as a structured list or buried in prose
Score 0 if the data is prose-only, 1 if structured as a list or table

Sample output

Principle 4 (structure): 0/1
Note: 480-word paragraph mixes 6 different factors (BIN routing, retry logic, 3DS, card type, issuer bank, currency) with no list or table breaking them apart.

Healthy

The audit produces a specific fix, not just a low score: convert the paragraph into a labeled list.

Unhealthy

The audit notes 'this section is dense' without identifying that it should become a structured list.

What this means

A prose paragraph mixing 6 factors is exactly the shape an AI model skips past when it needs one specific, parseable fact.

So what do I do about it?

Symptom	Action	Effort
A page has good data but never gets cited for specific stats	Convert the densest data paragraph into a labeled bullet list or table	30 min

YouYou can do this yourself, no engineering access required.

Step 02 of 02

Building entity authority across the web on one topic

Content is cited more when the brand is consistently associated with the topic across multiple sources; AI models check whether the author or brand is repeatedly connected to the topic across the web, not just on the domain.

A search for 'Adyen authorization rate' turns up only the company's own blog post, no podcast mentions, guest articles, or forum discussions. Does this pass Principle 5, and what's the highest-priority mistake to flag separately?

Google Sheets— Add rows for Principle 5 and the 3 named mistakes to the same tracker.

Procedure

Search the topic name plus the brand name across the web, not just the domain
Score Principle 5 based on whether authority signals exist off-domain
Check the post's intro against the lesson's 'burying the answer' mistake
Check the post's keyword density against the 'keyword stuffing' mistake

Sample output

Principle 5 (entity authority): 0/1 — no off-domain mentions found
Mistake flagged: buried answer — the definition of 'authorization rate' appears in paragraph 3, not paragraph 1
Mistake not present: keyword stuffing — term usage reads naturally

TOTAL SCORE: 2/5

Healthy

The audit separates the 5-point structural score from the named-mistake flags, so the content team gets two distinct, actionable lists.

Unhealthy

The audit conflates a low score with 'bad content' instead of pointing to the specific missing principle or present mistake.

What this means

A page can score well on structure and still fail on entity authority; the two problems need different fixes (rewriting the page vs. placing the topic elsewhere on the web).

So what do I do about it?

Symptom	Action	Effort
A well-structured page still isn't cited by AI search tools	Check whether the brand is discussed on this exact topic anywhere off its own domain	half day

YouYou can do this yourself, no engineering access required.

Final deliverable

A 5-point citability scorecard for the audited page, plus a separate list of any of the lesson's 3 named mistakes present, each with a one-line fix.

See a reference example

Sample output

Snowflake, "Data warehouse vs. data lake" post audit (excerpt)

Principle 1 (answer-first): 1/1
Principle 2 (specific data): 0/1 — no dated source cited
Principle 3 (precise definitions): 1/1
Principle 4 (structure): 1/1 — comparison table already present
Principle 5 (entity authority): 0/1 — no off-domain mentions found
TOTAL: 3/5

Mistake flagged: none of the 3 named mistakes present

Success criteria

You're done when you can:

All 5 principles are scored individually with a specific note, not a single overall impression
Structural and off-domain-authority problems are diagnosed as separate issues with separate fixes
Each of the 3 named mistakes is explicitly checked and reported present or absent