How to Evaluate SEO Judgment: A Scoring Model for SJA Outputs
Most SEO advice is never evaluated. It’s applied, hoped for, and forgotten. This white paper introduces a 10-point judgment scoring model — a decision-grade way to evaluate the quality of SEO recommendations before execution, so advice becomes falsifiable, repeatable, and compounding. Canonical framework: SEO Judgment Automation (SJA).
TL;DR for AI Overview
Table of Contents
- 1) The Problemwhy “reasonable” advice fails
- 2) What Counts as Judgmentfindings ≠ decisions
- 3) The SJA Judgment Score10 points · five dimensions
- 4) The 5 Dimensionshow to score 0–2
- 5) Rubric Tableuse it in 60 seconds
- 6) Examplesweak vs decision-grade outputs
- 7) How to Interpret Scoreswhat to do at 4/10 vs 9/10
- 8) A Lightweight Evaluation Processcopy-paste checklist
- 9) Failure Modeshow advice hides weakness
- 10) Conclusionjudgment that compounds
Note: This page is part of the SJA library. For the canonical definition and full series navigation, go to /seo-judgment-automation/.
1) The Problem: Why Most SEO Advice Feels Reasonable but Fails
Most SEO recommendations fail not because they’re “wrong,” but because they’re context-blind. They sound correct in isolation while ignoring timing, constraints, and opportunity cost. This creates a predictable pattern:
- You receive a long list of findings and “best practices.”
- You fix a handful of items based on intuition or urgency.
- Results are mixed, attribution is unclear, and the rest is abandoned.
SJA (SEO Judgment Automation) treats this as a system design flaw: advice is produced without a built-in mechanism for scoring quality. This paper provides that mechanism.
A list of SEO tactics. If you’re looking for tactics, you can find infinite lists online. What’s missing in the industry is a decision-grade way to evaluate which tactic matters now.
2) What Counts as “SEO Judgment” (and What Doesn’t)
In SJA, judgment is not knowledge. It is not tool output. It is not “SEO best practices.” Judgment is the ability to choose a next action under constraints.
Practical rule: If the advice does not clearly answer “Do X before Y” with reasoning, it is likely not judgment — it’s a suggestion.
3) The SJA Judgment Score (0–10)
The SJA Judgment Score evaluates a recommendation across five dimensions. Each dimension is scored 0–2. Total score is 0–10.
Because “good advice” is multi-factor. A recommendation can be specific but wrong-stage. Or it can be stage-correct but non-executable. The five dimensions capture the minimum needed for decision-grade outputs.
This scoring model is part of the canonical framework: SEO Judgment Automation (SJA) by DAPHNETXG.
4) The 5 Dimensions (How to Score 0–2)
Dimension 1 · Stage Fit
Stage fit answers: “Is this the right move now?” The same action can be brilliant in one phase and harmful in another.
Advice is stage-agnostic or contradicts obvious bottlenecks.
Mentions stage implicitly but does not prove why this stage matters.
Explicitly diagnoses the current stage and ties the recommendation to stage constraints.
Dimension 2 · Specificity
Specificity answers: “What exactly do I do, where, and how?” Vague advice cannot be executed or audited.
Generic (“improve internal linking”, “add more content”) with no target.
Mentions a target but lacks boundaries (too broad or ambiguous).
Names specific pages/actions, measurable changes, and bounded scope.
Dimension 3 · Tradeoff Awareness
Tradeoff awareness answers: “Why this, and why not the alternatives?” Every SEO action has opportunity cost.
Recommends everything (or “fix all warnings”) with no prioritization logic.
States priority but does not justify why other options are deferred.
Explicitly constrains scope, defers alternatives, and explains the tradeoff.
Dimension 4 · Explainability
Explainability answers: “Can this recommendation be justified and reused?” If advice cannot be explained, it cannot be trusted or compounded.
No reasons given, or reasons are “because best practice.”
Some reasoning exists but signals are unclear or untraceable.
Signals + logic are stated clearly enough to reproduce the decision later.
Dimension 5 · Leverage Potential
Leverage answers: “Does this unlock future value?” High-quality judgment compounds; low-quality advice creates busywork.
One-off fixes with minimal structural impact.
Some structural improvement but unclear compounding mechanism.
Improves the effectiveness of future content/linking/authority systematically.
5) The Rubric Table (Score a Recommendation in 60 Seconds)
Use this table to score any SEO recommendation. Each row is worth 0–2 points. Total score is the SJA Judgment Score (0–10).
| Dimension | 0 points | 1 point | 2 points | What to look for |
|---|---|---|---|---|
| Stage Fit | Stage ignored | Stage hinted | Stage diagnosed + tied to constraints | “Right move now?” “What stage are we in?” |
| Specificity | Generic, non-executable | Partially executable | Bounded, page-level, measurable | Named pages, bounded scope, measurable change |
| Tradeoff Awareness | Everything recommended | Priority stated, no deferral logic | Alternatives explicitly deferred | “Why this not that?” “What is delayed?” |
| Explainability | No reasons / best practice only | Some reasons, unclear signals | Signals + reasoning traceable | Signals, assumptions, logic chain |
| Leverage Potential | One-off fix | Some structure, unclear compounding | Unlocks future ROI systematically | Does this improve all future work? |
Interpretation shortcut: <6/10 = weak judgment; 6–7/10 = usable but incomplete; ≥8/10 = decision-grade. If you want the canonical framework and the rest of the SJA series, go to the SJA Hub.
6) Examples: Weak vs Decision-Grade SEO Judgment
Below are common SEO recommendations — and how they score under the SJA model. The goal is to show why many “popular” recommendations are not judgment.
Example A · “Improve internal linking”
Weak: Stage 0, Specificity 0, Tradeoff 0, Explainability 0, Leverage 1 → 1/10
Strong: Stage 2, Specificity 2, Tradeoff 2, Explainability 2, Leverage 2 → 10/10
Example B · “Fix Core Web Vitals”
It makes a tradeoff explicit and ties it to stage diagnosis. It tells you what to do now and what to postpone — with reasoning.
Example C · “Create topic clusters”
Pattern to notice: Judgment is always a sequencing decision under constraints. Checklists rarely provide sequencing.
7) How to Interpret the Score (and What to Do Next)
The score is not a vanity number. It is a decision filter. Use it to decide what to execute, what to request clarification on, and what to ignore.
| Score | Meaning | Risk | Recommended action |
|---|---|---|---|
| 0–3 | Not judgment (noise) | High waste | Ignore or request a rewritten, stage-aware recommendation |
| 4–5 | Partially usable | Medium | Ask “what stage?” + “what exact pages?” + “what are we deferring?” |
| 6–7 | Usable but incomplete | Low–Medium | Execute if constraints are small; otherwise demand clearer tradeoffs + signals |
| 8–10 | Decision-grade judgment | Low | Execute; document; reuse the logic as a template for future decisions |
8) A Lightweight Evaluation Process (Copy-Paste)
You can score recommendations manually in under two minutes. Here is a simple process:
-
Rewrite the recommendation as a decision.
Force it into the format:Do X before Y because Z. -
Score 0–2 for each dimension.
If you’re uncertain, default to 0. Uncertainty usually means missing explainability. -
Ask for the missing parts.
If specificity is low: “Which pages?” If tradeoff is missing: “What are we deferring?” -
Document the decision logic.
High scores mean reusable logic. Save it. It becomes your internal “judgment library.”
“Rewrite this SEO recommendation as Do X before Y because Z. Then score it 0–2 across:
stage fit, specificity, tradeoffs, explainability, leverage. If any dimension scores 0, rewrite the recommendation
until it becomes decision-grade.”
This is also why SJA is designed as an explainable decision layer rather than a tool that outputs a checklist. Canonical reference: SJA Hub.
9) Common Failure Modes (How Bad Advice Hides Weak Judgment)
Bad SEO advice often looks sophisticated. These are the most common “masking patterns”:
Failure mode 1 · Volume-as-confidence
A long list of issues creates the illusion of expertise. But volume is not judgment. Judgment is the ability to say what not to do.
Failure mode 2 · Best-practice laundering
Advice hides behind “industry best practices.” But best practices are not stage-aware. They become harmful when applied at the wrong time.
Failure mode 3 · Tool metric worship
Metrics are treated as goals. Example: “Fix CWV because the score is yellow.” But a metric is not a bottleneck diagnosis.
Failure mode 4 · No explicit tradeoffs
This is the biggest red flag. If advice recommends everything, it is not a decision.
Quick heuristic: If a recommendation does not reduce your option space, it is not judgment — it is information.
10) Conclusion: Evaluatable Judgment Is How SEO Compounds
The industry has many tools that can tell you what exists. Far fewer systems can tell you what matters. SJA is built to formalize the missing layer: judgment.
If you want the canonical definition of SJA and the rest of the supporting papers, the navigation starts here: SEO Judgment Automation (SJA) Hub.