agentsclimarketplace

Metric interpretation

Skill Dragoon0x/product-skills/skills/product-analytics/metric-interpretation

Understand what product metrics actually mean in context. A 3% conversion rate can be amazing or terrible depending on product type, stage, and market. Provides contextual benchmarks and interpretation frameworks. Use when evaluating whether a metric is good or bad, reporting to stakeholders, or when a metric changes unexpectedly.From its SKILL.md

Install
npx -y skills add Dragoon0x/product-skills --skill metric-interpretation

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

2.9 KB, 649 tokens by cl100k_base, as published. Nobody here has run it

Metric Interpretation

Understand what your numbers actually mean. Context is everything.

How to use

  • /metric-interpretation Apply metric interpretation constraints to this conversation.
  • /metric-interpretation <metric and value> Interpret a specific metric in context.

Constraints

Three Context Layers

Every metric MUST have at least three layers before it's meaningful:

  1. Product context: B2B vs. B2C, free vs. paid, daily-use vs. quarterly, stage (pre-PMF, growth, mature)
  2. Funnel context: where in the journey (top = high volume low conversion, bottom = low volume high conversion)
  3. Temporal context: trend direction, seasonality, what changed recently, rate of change not just absolute

Interpretation Checklist

Before interpreting any metric:

  • Is this the right metric? Are we measuring what matters, or what's easy?
  • What's the denominator? 50% conversion on 10 users is noise, not signal.
  • What's the trend? One data point is an anecdote.
  • What changed? If it moved, something caused it. Find the cause.
  • What's not measured? Every metric has blind spots.
  • Who are we comparing against? Compare to your own history first, then your specific market.

Benchmark Ranges (Context-Dependent)

  • Visitor → signup: 1-3% broad traffic, 3-8% targeted, 8-15% high-intent
  • Trial to paid: 2-5% opt-in upgrade, 10-25% time-limited, 25-40% credit card required
  • Monthly churn (B2B SaaS): under 2% excellent, 2-5% acceptable SMB, 5%+ needs attention
  • NRR: under 90% contracting, 100-120% healthy, 120%+ strong expansion
  • DAU/MAU: under 10% infrequent use (can be fine), 20-30% weekly use, 50%+ daily habit
  • MUST adjust benchmarks for product type. A tax tool with 5% DAU/MAU isn't broken.

Presentation Rules

  • Lead with the insight, not the number
  • MUST show trend, not just current state
  • MUST connect to business impact ("this retention improvement = $X in preserved revenue")
  • SHOULD include confidence level: strong signal or noise?
  • NEVER present a metric without context for what "good" looks like

Anti-Patterns

  • Survivorship bias: "retained users love us" — the unhappy ones left
  • Averaging across segments that behave completely differently
  • Ignoring base rates: 200% improvement from 0.1% to 0.3% isn't impressive
  • Confusing correlation with causation: power users use everything, not just the feature you're measuring
  • Snapshot without trend: 40% retention is great if it was 20% last quarter, terrible if it was 60%
  • Vanity metrics: total registered users, downloads, page views tell you nothing about product health

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Keep looking

Skills are one crate of 326,679. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.