Artifact-to-Tactic Context Fit Rule
Match the evidence you provide to the marketing tactic you want graded
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 6
- Confidence
- 95%
This decision rule separates grader failure from context failure. Before accepting an AI evaluation, compare the requested tactic with the artifacts actually supplied. A product-launch audit needs launch collateral, sequencing, audience information, offers, demonstrations, proof, and distribution context; a landing page alone mainly exposes copy and positioning. Likewise, SEO work, paid campaigns, and sales calls require their own representative evidence. If the resulting principles or recommendations drift toward a neighboring discipline, first inspect the inputs rather than immediately rewriting the prompt. Add the missing context, clarify the audience and objective, and rerun the evaluation. Only revise the grader's logic if the assessment remains misaligned after the evidence properly represents the task. The rule improves both diagnostic validity and the usefulness of recommended actions.
Origin
The host asked the universal grader to evaluate a product launch but supplied a landing page, then recognized that the output skewed toward positioning because the launch assets were missing. Extracted from Marketing Against The Grain.
Core principles
- 01An evaluator can only judge the context it receives
- 02Input artifacts must represent the requested tactic
- 03Audience and objective improve the relevance of the assessment
- 04Misaligned evidence shifts the grader toward a neighboring discipline
- 05Unexpected output should trigger an input audit before a prompt rewrite
How to run it
- 1
Define the Evaluation Target
State the exact tactic, audience, objective, and decision the evaluation should support.
Pro tip Describe what success for this evaluation would enable you to decide or improve.
Watch out A tactic name without an objective may still permit several incompatible interpretations.
- 2
Map Required Evidence
List the artifacts and contextual facts a competent human evaluator would need to assess that tactic.
Pro tip Include process evidence and outcomes, not just the most visible final asset.
Watch out Do not assume one landing page represents an entire campaign or launch.
- 3
Supply Representative Inputs
Provide the complete or best available evidence set in a form the AI can interpret.
Pro tip Sales-call analysis can use transcripts, while SEO analysis can use audits, queries, pages, and performance data.
Watch out Missing materials create blind spots that confident prose can conceal.
- 4
Inspect Evaluation Alignment
Check whether the generated principles, scores, and recommendations actually belong to the requested tactic.
Pro tip Flag criteria that repeatedly describe an adjacent discipline such as positioning when the target is launch execution.
Watch out Useful-looking recommendations can still answer the wrong question.
- 5
Correct Context Before Logic
When the audit drifts, add missing artifacts or clarify the audience and objective before modifying the grader's methodology.
Pro tip Tell the grader which supplied asset serves which part of the tactic.
Watch out Changing prompt logic to compensate for incomplete evidence can make the grader brittle.
- 6
Rerun and Compare
Repeat the evaluation and compare whether its principles and priorities now cover the intended tactic more completely.
Pro tip Retain the earlier output as a diagnostic baseline.
Watch out If drift persists with representative inputs, the expert selection or principle-generation instructions may need revision.
In the wild
The host requests a product-launch evaluation but submits a CRM landing page instead of the complete launch assets. The grader produces some launch advice, such as a product animation and old-way/new-way narrative, but much of the assessment remains focused on copy and positioning. The host identifies the missing context as the reason for the skew.
→ The mismatch demonstrates that the requested tactic and supplied evidence must align before the grade can be trusted.
Common mistakes
Blaming the grader first
Misaligned output may be a faithful response to incomplete or unrepresentative inputs rather than a defect in the prompt.
Supplying only the easiest asset
A convenient page or document may reveal only one component of a broader marketing system.
Omitting the audience
Without the intended buyer, otherwise sound criteria and recommendations may optimize for the wrong market segment.
Is it for you?
Best for
Anyone using AI to grade campaigns, launches, positioning, SEO work, sales calls, or multi-asset marketing systems.
Not ideal for
Simple self-contained tasks where one artifact fully represents both the tactic and its context.
From the transcript
“Context really matters.”
“Again, context really matters, and because it's the page, it's really honing in on positioning versus the launch because I haven't given it the launch…”
“as long as you give it the context to actually evaluate.”
From the episode
I Built a Personal Marketing Coach With One Prompt