campaign-analytics

Files (4)

• SKILL.md
• framework-notes.md · references/
• check-output.py · scripts/
• output-template.md · templates/

SKILL.md (first lines)

# Campaign Analytics

## Overview

Campaign Analytics provides a systematic framework for measuring, diagnosing, and optimizing outbound and inbound campaign performance. The core principle is that campaigns fail at specific layers in the infrastructure-to-revenue stack — fixing the wrong layer wastes time and money. This skill prevents the common mistake of tweaking copy when deliverability is broken, or adjusting targeting when the offer is wrong.

The non-obvious rule: every campaign metric exists in a dependency chain. Deliverability gates open rates. Open rates gate reply rates. Reply rates gate meetings. Meetings gate pipeline. You cannot fix layer 5 if layer 2 is broken. This skill enforces layer-by-layer diagnosis before optimization.

This skill produces a Campaign Performance Diagnostic Report containing: a 6-layer health scan, a diagnostic decision tree with root cause identification, winning copy extraction with statistical confidence, benchmark comparisons across three reference sets, and a prioritized optimization action plan with expected impact estimates.

## When to Use

- User says "analyze my campaign" or "campaign performance review" → activate this skill
- User mentions "open rates dropped" or "reply rates are low" → use this skill for root-cause diagnosis
- User asks "which subject line performed better" or "what copy is working" → use the copy extraction module
- User says "benchmark my metrics" or "how do I compare to industry" → activate benchmarking
- User mentions "campaign ROI" or "cost per meeting" → use full stack analysis
- User asks "why isn't this campaign working" → run the diagnostic decision tree
- Trigger phrases: campaign analytics, email performance, sequence metrics, campaign optimization, campaign dashboard, pipeline from campaigns

Do NOT use for:
- Setting up new campaigns or sequences → use cold-email-strategy
- Writing new email copy → use cold-email-copywriting
- Deliverability-only issues (DNS, warmup, blacklists) → use deliverability-monitoring
- Attribution modeling across channels → use attribution
- Real-time alerting on pipeline metrics → use proactive-alerts

## Authoritative Foundations

This skill draws from the following established methodologies:

- **Winning by Design — Bowtie Model** — Revenue visualization that extends beyond the funnel through retention and expansion. Campaign analytics map to the acquisition side of the bowtie: awareness → engagement → opportunity → close. Each transition has measurable conversion rates that this skill diagnoses.

- **Winning by Design — GTM Index** — 1-10 scoring across 6 models (Revenue, Data, Math, Operating, Growth, GTM). The Data model specifically governs measurement discipline. Campaign analytics without reliable data architecture produces misleading conclusions. This skill applies Data Model scoring to campaign measurement infrastructure.

- **David Skok — SaaS Metrics** — Pipeline velocity and conversion rate analysis. Skok's work on understanding sales funnel dynamics through cohort analysis informs the layer-by-layer diagnostic approach. Campaigns produce cohorts; comparing cohort performance reveals what's changing over time.

- **Gartner/CEB — Challenger Sale** — Insight-led selling. Campaign effectiveness correlates with the quality of commercial insight delivered. This skill's copy extraction module evaluates whether winning messages teach, tailor, and take control versus simply pitching.

## Prerequisites

- Campaign data in structured format (CSV, Google Sheets, or CRM export) with at minimum: send date, opens, clicks, replies, meetings booked, opportunities created, pipeline value, and won deals — all keyed by campaign/variant
- Access to sending platform analytics (Smartlead, Instantly, Salesforge, Apollo, or Outreach)
- CRM access for pipeline attribution (HubSpot, Salesforce, or Attio)
- Optional: deliverability data from tools like MXToolbox, Google Postmaster, or Validity for layer 1 diagnosis
- Optional: Google Analytics or attribution platform data for inbound campaign analysis
- Recommended upstream skill: deliverability-monitoring (if bounce rates exceed 2%)

## Step-by-Step Process

### Phase 1: Intake

Gather required information from the user. Ask all questions at once. Do not proceed until all answers are received.

Required intake questions:

1. **Campaign scope**: Which campaigns, sequences, or time periods should be analyzed? Define start and end dates.
2. **Data sources**: Where is campaign data stored? (Smartlead, Instantly, CRM, spreadsheet?) Request export or access.
3. **Goals**: What were the campaign objectives? (Meetings booked, pipeline generated, revenue closed?) What targets were set?
4. **Segments**: Were there audience segments, ICP tiers, or personas targeted? List them for cohort analysis.
5. **Variants tested**: Were A/B tests run? If so, what was varied (subject lines, copy, offer, timing, channel)?
6. **Budget context**: Total spend including tools, data enrichment, and team time. Needed for cost-per-meeting and ROI calculations.
7. **Benchmark context**: Company stage (seed, Series A, growth), ACV range, ICP type (SMB, mid-market, enterprise), and industry vertical for appropriate benchmark selection.

### Phase 2: Data Assembly

Load and validate campaign data before analysis. This phase ensures data integrity.

1. **Import all data sources** into a unified analysis structure. Map fields to a canonical schema: campaign_id, variant_id, send_date, recipient_count, delivered, opened, clicked, replied, bounce_type, meeting_booked, meeting_date, opportunity_created, opportunity_value, deal_closed, deal_value, segment, persona, channel.

2. **Validate data integrity**:
   - Check for missing data in critical fields (delivered, opened, replied)
   - Verify send date consistency — no future dates, gaps explained
   - Confirm recipient counts match between sending platform and CRM
   - Flag any campaigns where delivered count < 50% of sent (possible infrastructure issue)
   - Test reply attribution: do meeting_booked counts align with replied-to-meeting conversion rates that make sense?

3. **Calculate raw metrics** for every campaign/variant:
   - Delivery Rate = delivered ÷ sent
   - Open Rate = unique opens ÷ delivered
   - Click Rate = unique clicks ÷ delivered
   - Reply Rate = replies ÷ delivered
   - Meeting Rate = meetings_booked ÷ delivered
   - Opportunity Rate = opportunities ÷ meetings_booked
   - Pipeline Rate = total_pipeline_value ÷ opportunities
   - Win Rate = deals_closed ÷ opportunities
   - Cost Per Meeting = total_campaign_cost ÷ meetings_booked
   - Cost Per Opportunity = total_campaign_cost ÷ opportunities_created

4. **Segment data** by cohort dimensions: ICP tier, persona, channel, time period (weekly cohorts for trend analysis), and variant (for A/B analysis).

### Phase 3: 6-Layer Diagnostic Stack

Run the complete 6-layer diagnostic. Each layer must be healthy before diagnosing the next.

**Layer 1: Infrastructure & Deliverability**

Metrics checked: Delivery rate, bounce rate by type (hard vs soft), spam complaint rate, inbox placement rate.

Diagnostic thresholds:
- Delivery rate < 95% → critical. Check DNS (SPF/DKIM/DMARC), blacklist status, sending reputation.
- Hard bounce rate > 2% → critical. Pause campaign. Data source or verification process is failing.
- Soft bounce rate > 3% → warning. Possible mailbox full, temporary server issues, rate limiting.
- Spam complaint rate > 0.1% → critical. Content or targeting problem. Google/Microsoft thresholds are 0.1-0.3%.

Only proceed to Layer 2 if Layer 1 passes all thresholds.

**Layer 2: Audience & Targeting**

Metrics checked: Open rate, open-to-reply rate, unsubscribe rate, negative reply sentiment ratio.

Diagnostic thresholds:
- Open rate < 30% (cold) or < 50% (warm) → audience mismatch or subject line failure. Check: is ICP definition accurate? Are contacts still at company? Are job titles current?
- Unsubscribe rate > 0.5% → targeting or frequency problem. Audience doesn't want this message.
- Negative reply sentiment > 20% of all replies → messaging-audience mismatch. Wrong persona or wrong problem.

Decision tree: If open rate is low AND Layer 1 is healthy → subject line or sender reputation problem. If open rate is healthy but reply rate is low → copy or offer problem (diagnose at Layer 3).

**Layer 3: Messaging & Copy**

From LeadMagic/gtm-skills, MIT.

More from LeadMagic/gtm-skills