BUYER'S GUIDE

Do not start with the agency's tactics. Start with what it can prove.

A credible AEO partner should define what it measures, preserve the underlying evidence, disclose uncertainty, and separate observed visibility from claimed causation.

If the proposal cannot do that, the terminology does not matter.

proposal anatomy4 layers + 1 flag
01Baseline
02Evidence
03Interpretation
04Action
!Guaranteesdetached · treat with scrutiny
the eight questions

Ask these before you sign.

1
What exactly will you measure?

Look for defined prompts, platforms, markets, dates, mentions, citations, competitors, accuracy, and a repeatable observation method.

2
Will we see the underlying answers?

A score without the answer, source, date, and prompt is difficult to review or challenge.

3
How do you account for answer variability?

One prompt on one platform is an observation. A baseline requires controlled repetition and clearly stated coverage.

4
How do you separate evidence from inference?

The partner should distinguish what appeared from why it may have appeared. A visibility change does not prove that one optimization caused it.

5
What work is actually AEO-specific?

Technical SEO, content, schema, digital PR, reviews, and entity consistency may all contribute. The proposal should explain how each activity connects to an observed answer gap.

6
What can you not promise?

No agency controls a model's output. Guaranteed citations, mentions, or rankings should trigger scrutiny.

7
What will we receive?

Require reviewable findings, source records, limitations, prioritized actions, ownership, and a method for re-measurement.

8
Who owns the recommendation?

Automation can collect and organize observations. A named analyst should be accountable for interpretation, caveats, and recommendations.

repackaged seo or connected intelligence?

Tell the two proposals apart.

WEAK PROPOSAL
Begins with a tactic list
Treats one screenshot as proof
Sells schema as the answer
Promises rankings or citations
Reports one blended score
Hides the method
STRONGER PROPOSAL
Begins with a documented baseline
Preserves repeated observations
Explains what schema can and cannot clarify
States limits and testable hypotheses
Separates mentions, citations, accuracy, and competitors
Makes scope and method reviewable

SEO work may be part of the remedy. The problem is not overlap. The problem is relabeling familiar work without connecting it to an observed AI visibility gap.

Read AEO vs. SEO →
what proof should look like

Proof is a record, not a promise.

Case studies can show prior work. They still should not be presented as a guarantee that another brand, market, or model will behave the same way.

Learn how AI visibility is measured →
evidence recordIllustrative
Prompt & market
"best signage co · Austin"
Platform & date
[Platform] / [Date]
Preserved answer
Full text stored
Brand mentioned?
No
Source cited?
Directory
Competitors present
A, C
Description accurate?
n/a — absent
Analyst interpretation
Weaker local trail
Stated limitation
Directional until repeated
human accountability

The useful question is not whether an agency uses AI. It is whether a responsible person can explain and defend the work.

evaluation scorecard

Compare partners on what serious work must disclose.

Evaluation area
What good looks like
Scope
Prompts, platforms, markets, dates, and exclusions are defined
Evidence
Original answers and cited sources are preserved
Measurement
Metrics remain distinct and repeatable
Interpretation
Observation is separated from inference
Remedies
Actions connect to documented gaps
Limitations
Unknowns and platform variability are disclosed
Accountability
A named analyst owns the conclusion
check each area against the proposal in front of you · no invented total score
cited by bnarrativ

Establish the baseline before buying the remedy.

Cited by Bnarrativ records what AI systems say, which sources they cite, which competitors appear, and where the evidence remains incomplete.

Explore Cited by Bnarrativ →
related reading
01
What Is AEO?
03
How to Measure AI Visibility
04
What Makes a Brand Citable?