METRICS CATALOG
> DEFINING WHAT TO MEASURE // THE 0-TO-1 ANALYTICS ENGINEERING JOURNEY
Metrics are not just numbers — they are decisions waiting to happen. Each metric below was chosen because it answers a specific business question that a stakeholder will ask. The formula tells you how to compute it. The threshold tells you when to act. The "why it matters" tells you what decision it informs.
FORMULA
THRESHOLDS
<70% — investigate demand gap
80-90% — healthy range
>95% — supply constrained, raise floor prices
WHY IT MATTERS
Are we monetizing enough of our available inventory? Every unfilled ad slot is lost revenue that can never be recovered — ad inventory is the most perishable asset in media.
FORMULA
THRESHOLDS
<$8 — below market, check demand quality
$12-$25 — healthy streaming range
>$30 — premium inventory signal
WHY IT MATTERS
What is our inventory worth? eCPM is the universal pricing currency for ad inventory. It normalizes revenue across different deal types (CPM, CPC, CPA) into a single comparable value.
FORMULA
BENCHMARKS BY SLOT
Pre-roll — ~92% (captive audience)
Mid-roll — ~85% (content investment)
Post-roll — ~55% (low motivation to stay)
WHY IT MATTERS
Are viewers watching our ads? Completion rate directly impacts advertiser satisfaction and renewal rates. Low completion signals ad fatigue, poor targeting, or excessive ad load.
FORMULA
IAB: 50% pixels visible for 2 continuous seconds
THRESHOLDS
<60% — below industry standard
~70% — industry benchmark
>80% — premium, charge accordingly
WHY IT MATTERS
Can advertisers trust our inventory? Viewability is the baseline credibility metric. Advertisers increasingly demand viewable-only billing — if your viewability is low, your effective yield collapses.
FORMULA
THRESHOLDS
<$2/hr — under-monetized
~$4/hr — target equilibrium
>$6/hr — risk of ad overload, monitor churn
WHY IT MATTERS
THE metric that balances monetization against user experience. RPSH is the single number that tells you if your ad load is sustainable. Push it too high and subscribers churn. Too low and the business model fails.
FORMULA
THRESHOLDS
<0.9 — under-pacing, at risk of under-delivery
0.9-1.1 — on pace
>1.1 — over-pacing, throttle or risk budget exhaustion
WHY IT MATTERS
Is the campaign spending on schedule? Under-delivery means the advertiser doesn't get what they paid for. Over-delivery means budget exhaustion before the flight ends. Both erode trust.
FORMULA
THRESHOLDS
<2% — target compliance
2-5% — investigate ad server config
>5% — ad fatigue risk, immediate action
WHY IT MATTERS
Are we annoying viewers? Frequency capping protects user experience. When the same ad repeats excessively, viewers associate the negative experience with both the advertiser and the platform.
FORMULA
Herfindahl-Hirschman Index
THRESHOLDS
<1500 — diversified revenue base
1500-2500 — moderate concentration
>2500 — concentrated, single-advertiser risk
WHY IT MATTERS
Revenue diversification is survival. If one advertiser represents 40% of revenue, losing that account is existential. HHI quantifies this concentration risk using the same index the DOJ uses for antitrust analysis.
Metrics do not exist in isolation. Understanding their interactions is what separates a metrics catalog from actual analytics engineering:
HIGH FILL RATE + LOW eCPM
Selling cheap inventory. Floor prices too low, or demand quality is poor. Revenue opportunity is being left on the table.
LOW COMPLETION + HIGH FREQUENCY
Ad fatigue signal. Viewers are being shown too many ads they don't want to watch. Churn risk is elevated.
HIGH RPSH + DECLINING VIEWABILITY
Maximizing short-term revenue at the cost of ad quality. Advertisers will demand refunds or walk.
HIGH HHI + OVER-PACING TOP CAMPAIGNS
Concentrated revenue burning too fast. When those campaigns exhaust budget, fill rate and revenue will cliff.
The metrics catalog above covers platform health and operational monitoring. The next layer of analytics maturity would add:
- > Attribution modeling — Multi-touch attribution to connect ad exposure to subscriber conversion. Last-touch is table stakes; data-driven attribution is the goal.
- > Incrementality testing — Holdout experiments to measure true ad lift vs. organic behavior. Without incrementality, you're measuring correlation and calling it causation.
- > LTV-based bid optimization — Weight auction bids by predicted subscriber lifetime value. A high-LTV viewer seeing a well-targeted ad is worth more than raw impression volume.