Getting Started

Understanding LLM Scores

Learn how scores are calculated and what they mean for your optimization strategy.

Score Categories

Each category pinpoints a different optimization lever.

Relevance

How closely the content matches the target query and user intent.

Clarity

How easy it is for LLMs to extract the main answer and supporting points.

Coverage

Whether essential entities, topics, and definitions are included.

Authority

Signals that demonstrate expertise, sources, and trust indicators.

Structure

Use of headings, lists, and formatting that improves model parsing.

How Scoring Works

Use these notes to interpret scores accurately.

●Scores are normalized from 0 to 100

●Each category is weighted based on intent type

●Historical scores show progress over time

●Recommendations prioritize highest impact changes

Score Interpretation Framework

Use score trends and category movement to drive better decisions.

●Monitor category-level changes after each content deployment.

●Correlate score shifts with traffic and conversion outcomes.

●Investigate drops by page type rather than isolated pages only.

●Set threshold alerts for priority URLs to trigger audits quickly.

Execution Phases

Week 1: Foundation

Finish installation and validate data flow.

Output: Connected tracker, imported sitemap, baseline analysis report.

Week 2: Prioritization

Select high-impact pages and define task backlog.

Output: Prioritized optimization list with owners and due dates.

Week 3-4: Iteration

Ship updates and verify score + business impact.

Output: Before/after snapshots with measurable deltas.

Common Mistakes

●Starting automation before validating tracker and sitemap integrity.

●Treating overall score as the only KPI and ignoring category gaps.

●Rolling out broad template changes without controlled testing.

●Skipping documentation of changes, making results hard to attribute.

Success Metrics

Analyzed Coverage

>= 90% priority URLs analyzed monthly

Ensures optimization decisions are based on current data.

Score Movement

+8 to +15 points on target templates in 30-60 days

Validates that executed recommendations are improving quality.

Execution Velocity

At least 1 improvement cycle per week

Creates compounding gains from consistent iteration.