marketing-analytics

Imported from okou-ai/vm0-skills (marketing-analytics/SKILL.md). Install upstream with npx skills add okou-ai/vm0-skills --skill marketing-analytics. Copyright stays with the author.

Channel Metric Reference

Email

Social Platforms

Paid Advertising (Search and Social)

Organic Search / SEO

Content Performance

Pipeline and Revenue Metrics

Report Structures

Weekly Snapshot

Designed for rapid team consumption:

• Three headline metrics with week-over-week movement
• Wins: 1-2 data-backed highlights
• Watch items: 1-2 areas requiring attention with supporting numbers
• Upcoming actions: 3-5 priorities for the week ahead

Monthly Performance Review

Standard format for stakeholder reporting:

1. Executive summary (3-5 sentences)
2. Core metrics table with month-over-month and target comparisons
3. Channel-level performance breakdown
4. Campaign results and highlights
5. What succeeded and what underperformed, with working hypotheses
6. Recommendations and priorities for the coming month
7. Budget spent vs. planned

Quarterly Strategic Review

For leadership-level analysis:

1. Quarter results against stated goals
2. Year-to-date progress and trajectory
3. Channel-by-channel ROI assessment
4. Campaign portfolio performance summary
5. Competitive and market landscape observations
6. Strategic recommendations for the next quarter
7. Budget proposal and reallocation plan
8. Experiment outcomes and key learnings

Dashboard Construction Principles

• Feature the metrics that tie directly to business goals, not vanity numbers
• Display trends over multiple periods rather than isolated data points
• Provide comparison anchors: prior period, target, industry benchmark
• Apply uniform color signaling: green for on-track, yellow for at-risk, red for off-track
• Organize by funnel stage or the business question being answered
• Confine the dashboard to a single screen; relegate granular data to an appendix
• Match the refresh cadence to the decision cadence (real-time for paid media, weekly for content)

Trend Analysis and Projection

Spotting Patterns

When examining performance data, investigate:

1. Sustained direction: is the metric consistently rising, falling, or flat across 4+ consecutive periods?
2. Turning points: at what moment did the trajectory change, and what event coincided?
3. Cyclical patterns: are there recurring fluctuations by day of week, month, or quarter?
4. Outliers: isolated spikes or dips — what triggered them, and could the cause be replicated or avoided?
5. Predictive signals: which metrics shift first and foreshadow downstream outcomes?

Analytical Process

1. Plot the metric across time with at least 8-12 data points for statistical relevance
2. Characterize the overall trajectory (rising, declining, stable, or oscillating)
3. Quantify the rate of change — is the trend accelerating or flattening?
4. Layer in external events (campaign launches, product updates, market shifts)
5. Benchmark against targets or industry norms
6. Look for correlations with related metrics
7. Formulate causal hypotheses and design experiments to test them

Projection Techniques

• Trend extension: project the existing trajectory forward (works best for stable metrics)
• Rolling average: average the most recent 3-6 periods to dampen noise
• Year-over-year overlay: use the prior year's seasonal pattern, adjusted for a growth coefficient
• Funnel arithmetic: forecast outputs from inputs (X leads at Y% conversion rate yields Z customers)
• Scenario planning: model optimistic, expected, and pessimistic cases

Projection Guardrails

• Near-term forecasts (1-3 months) carry far more reliability than long-range ones
• Projections built on fewer than 12 data points should be labeled low-confidence
• External disruptions (market shifts, competitive moves, economic changes) can invalidate trend-based models
• Always express forecasts as ranges rather than single numbers

Attribution Fundamentals

Why Attribution Matters

Buyers rarely convert after a single interaction. Attribution assigns credit across the multiple touchpoints that precede a conversion, informing channel investment decisions.

Standard Attribution Models

Practical Attribution Advice

• If you have no attribution system, begin with last-interaction — it is the simplest and most immediately actionable
• Contrast first-interaction and last-interaction views to learn which channels drive discovery vs. closure
• Position-based (40/20/40) is a pragmatic default for most B2B organizations
• Algorithmic models need high conversion volumes to produce statistically sound results
• Treat attribution as directional intelligence, never as absolute truth
• Any multi-touch model is more informative than a single-touch model, and any model outperforms none

Attribution Traps

• Optimizing a single channel based on single-touch data can starve the rest of the funnel
• Awareness-oriented channels (display, organic social, PR) will consistently underperform in last-touch reports
• Conversion-oriented channels (branded search, retargeting) will consistently underperform in first-touch reports
• Self-reported attribution ("How did you hear about us?") offers useful qualitative signal but is unreliable for quantitative allocation
• Cross-device and cross-channel tracking gaps guarantee that attribution data is always incomplete

Optimization Methodology

Systematic Improvement Process

1. Detect: which metrics fall short of targets or benchmarks?
2. Locate: where in the funnel does the breakdown occur? (impressions, clicks, conversions, retention)
3. Theorize: what is causing the shortfall? (targeting, messaging, creative, offer design, timing, technical issues)
4. Rank: which interventions promise the greatest impact relative to effort?
5. Experiment: run a controlled test to validate or disprove the hypothesis
6. Evaluate: did the metric improve meaningfully?
7. Act: scale successful changes broadly; iterate on inconclusive or negative results

Intervention Levers by Funnel Position

Impact-Effort Prioritization

Score every optimization idea on two axes:

Impact (potential metric movement):

• High: directly addresses the primary bottleneck
• Medium: improves a contributing factor
• Low: yields incremental gains

Effort (implementation difficulty):

• Low: copy tweak, targeting adjustment, quick A/B test
• Medium: new creative asset, page redesign, workflow modification
• High: new tooling, cross-team initiative, major content production

Execution order:

1. High impact, low effort — execute immediately
2. High impact, high effort — plan and staff
3. Low impact, low effort — pursue if bandwidth allows
4. Low impact, high effort — defer or deprioritize

Experimentation Discipline

• Isolate a single variable per test for interpretable results
• Lock in the success metric before the test begins
• Calculate the required sample size in advance and resist ending tests prematurely
• Run each test for at least one complete business cycle (usually a full week for B2B)
• Record all experiments and outcomes, including negative and null results
• Circulate learnings across the team — a test that confirms the current approach still builds confidence

Ongoing Optimization Rhythm

• Daily: check paid campaign pacing, flag anomalies, review ad approval status
• Weekly: assess channel-level performance, pause lagging efforts, amplify winners
• Biweekly: rotate ad creative and launch new test variants
• Monthly: conduct a comprehensive performance review, surface new optimization opportunities, refresh projections
• Quarterly: reassess channel strategy, budget distribution, and audience targeting at a strategic level