Sequence Performance

Sequence Performance

Goes beyond vanity metrics. Most campaign reports tell you open rate and reply rate. This composite reads the actual emails you sent, reads every reply you received, classifies the responses, evaluates your copy, evaluates your lead quality, and tells you specifically what's working, what's not, and what to do about it.

Three layers of analysis:

1. Quantitative: The numbers — sends, opens, replies, bounces, conversions, by touch and by variant
2. Qualitative (Copy): Are the subject lines, email bodies, CTAs, and personalization actually good? What's landing and what's falling flat?
3. Qualitative (Replies): What are people actually saying? What objections keep coming up? What's generating genuine interest?

When to Auto-Load

Load this composite when:

• User says "how's my campaign doing", "sequence performance", "campaign review", "email analytics"
• User says "analyze my outreach", "why isn't my campaign working", "review my email results"
• An upstream workflow (Pipeline Ops, weekly review) triggers a campaign performance check
• A campaign has been running for 7+ days and has meaningful data

Step 0: Configuration (One-Time Setup)

On first run, collect and store these preferences. Skip on subsequent runs.

Outreach Tool Config

Your Company Context (for copy evaluation)

Benchmark Context

Store config in: clients/<client-name>/config/sequence-performance.json or equivalent.

Step 1: Pull Campaign Data

Purpose: Extract all campaign data: metrics, email copy, and reply content.

Input Contract

campaign_selection: {
  campaign_id: string | null        # Specific campaign ID (if known)
  campaign_name: string | null      # Campaign name to search for
  date_range: {                     # Optional — pull all data if not specified
    start_date: string
    end_date: string
  } | null
  include_replies: boolean          # Default: true — pull actual reply text
}
outreach_tool: string               # From config
access_method: string               # From config

Process

Pull three categories of data:

A) Campaign Metrics

By outreach tool:

B) Email Copy (Sequence Content)

Pull the actual email templates/copy for every touch in the sequence:

By outreach tool:

C) Reply Content

Pull the actual text of every reply received:

By outreach tool:

Data Standardization

Normalize into a standard structure:

campaign_data: {
  campaign_name: string
  campaign_id: string
  date_range: { start: string, end: string }
  status: "active" | "paused" | "completed"

  metrics: {
    total_sent: integer
    unique_recipients: integer
    total_opens: integer
    unique_opens: integer
    total_replies: integer
    bounces: integer
    unsubscribes: integer
    clicks: integer | null
    meetings_booked: integer | null
  }

  sequence: [
    {
      touch_number: integer
      delay_days: integer               # Days after previous touch
      variants: [
        {
          variant_id: string            # "A", "B", etc.
          subject: string
          body: string
          personalization_fields: string[]
          metrics: {
            sent: integer
            opens: integer
            replies: integer
            bounces: integer
          }
        }
      ]
    }
  ]

  replies: [
    {
      sender_name: string
      sender_title: string | null
      sender_company: string | null
      sender_email: string
      reply_text: string
      reply_date: string
      triggered_by_touch: integer
      triggered_by_variant: string | null
      existing_category: string | null  # Tool's categorization if any
    }
  ]

  lead_list_summary: {
    total_leads: integer
    by_title: [ { title: string, count: integer } ] | null
    by_industry: [ { industry: string, count: integer } ] | null
    by_source: [ { source: string, count: integer } ] | null
  } | null
}

Human Checkpoint

## Campaign Data Pulled

Campaign: [name]
Status: [active/paused/completed]
Date range: [start] to [end]
Sent: X emails to Y recipients
Replies received: Z (full text pulled for analysis)
Touches: N touches, M variants
Lead data available: [yes/partial/no]

Data looks complete? (Y/n)

Step 2: Quantitative Analysis

Purpose: Calculate all performance metrics, benchmark against industry standards, and identify statistical patterns. Pure computation.

Input Contract

campaign_data: { ... }                # From Step 1
outreach_type: string                 # From config ("cold" or "warm")
target_segment: string                # From config ("smb", "mid-market", "enterprise")

Benchmarks

Calibrate expectations based on outreach type and segment:

Metrics Calculation

A) Overall Campaign Metrics

B) Per-Touch Breakdown

For each touch in the sequence:

C) Variant Analysis (A/B Testing)

If multiple variants exist per touch:

Statistical significance check:

• <50 sends per variant → "Insufficient data — too early to call"
• 50-100 sends → "Directional — [variant] is leading but not conclusive"
• 100-250 sends → "Likely winner — [variant] outperforms by X%"
• 250+ sends → "Statistically significant — scale [variant], kill [other]"

D) Temporal Patterns

Output Contract

quantitative_analysis: {
  overall: {
    open_rate: percentage
    reply_rate: percentage
    positive_reply_rate: percentage
    bounce_rate: percentage
    unsubscribe_rate: percentage
    click_rate: percentage | null
    meeting_conversion: percentage | null
    deliverability_rate: percentage
  }
  vs_benchmark: {
    open_rate: "above" | "at" | "below"
    reply_rate: "above" | "at" | "below"
    bounce_rate: "healthy" | "concerning" | "critical"
    assessment: string                 # One-line summary
  }
  by_touch: [
    {
      touch_number: integer
      sent: integer
      open_rate: percentage
      reply_rate: percentage
      marginal_reply_rate: percentage
      reply_contribution: percentage   # What % of total replies came from this touch
    }
  ]
  by_variant: [
    {
      touch_number: integer
      variant_id: string
      subject: string                  # For quick reference
      sent: integer
      open_rate: percentage
      reply_rate: percentage
      confidence: string               # "insufficient", "directional", "likely", "significant"
      recommendation: string           # "scale", "keep testing", "kill"
    }
  ] | null
  temporal: {
    best_day: string
    best_time: string
    avg_open_to_reply_hours: float
    velocity_trend: "accelerating" | "steady" | "decelerating"
  } | null
}

No human checkpoint — feeds directly into the next analyses.

Step 3: Reply Analysis

Purpose: Read every reply, classify it, extract objection patterns, and assess what the replies tell us about copy and lead quality. Pure LLM reasoning.

Input Contract

replies: [...]                        # From Step 1 campaign_data.replies
your_company: {
  description: string
  pain_point: string
  icp_definition: string
}

Process

A) Reply Classification

Read each reply and classify into categories:

B) Objection Pattern Analysis

Group objections and identify patterns:

C) Positive Signal Analysis

For positive replies and warm responses:

D) Reply Quality Score

Aggregate assessment:

Output Contract

reply_analysis: {
  total_replies: integer
  classification: [
    { category: string, count: integer, percentage: percentage }
  ]
  reply_quality_score: "strong" | "mixed" | "weak" | "toxic"
  reply_quality_reasoning: string

  objection_patterns: {
    top_objection: { type: string, count: integer, example_quotes: string[] }
    objections_by_touch: [ { touch: integer, objection_type: string, count: integer } ]
    handleable_objections: [ { type: string, count: integer, suggested_handle: string } ]
    terminal_objections: [ { type: string, count: integer, implication: string } ]
  }

  positive_signals: {
    interest_triggers: [ { touch: integer, variant: string | null, pattern: string } ]
    common_responder_traits: string     # What positive responders have in common
    resonating_phrases: string[]        # Exact quotes from positive replies
    common_questions: string[]          # Questions warm leads are asking
  }

  notable_replies: [                    # 5-10 most instructive replies
    {
      category: string
      sender_info: string               # Title @ Company
      quote: string                     # Key excerpt
      insight: string                   # What this tells us
    }
  ]
}

Step 4: Copy Quality Assessment

Purpose: Evaluate the actual email copy — subject lines, body, personalization, CTA — against best practices and against what the reply data tells us. Pure LLM reasoning.

Input Contract

sequence: [...]                       # From Step 1 campaign_data.sequence (the actual emails)
quantitative_analysis: { ... }        # From Step 2 (which touches/variants perform)
reply_analysis: { ... }               # From Step 3 (what replies say about the copy)
your_company: {
  description: string
  pain_point: string
  proof_points: string[]
  cta_goal: string
}

Process

Evaluate each touch and variant across these dimensions:

A) Subject Line Analysis

For each subject line:

Subject line verdict per variant:

• Working: Above-benchmark open rate. Keep it.
• Underperforming: Below-benchmark open rate. Diagnose why and suggest alternatives.
• Spam-flagged: Very low open rate + high bounce rate. Likely hitting spam filters.

B) Email Body Analysis

For each email body:

C) Personalization Assessment

D) Sequence Architecture Assessment

Output Contract

copy_assessment: {
  overall_grade: "A" | "B" | "C" | "D" | "F"
  overall_summary: string              # 2-3 sentence assessment

  subject_lines: [
    {
      touch: integer
      variant: string
      subject: string
      open_rate: percentage
      verdict: "working" | "underperforming" | "spam_risk"
      issues: string[]
      suggested_alternative: string | null
    }
  ]

  email_bodies: [
    {
      touch: integer
      variant: string
      word_count: integer
      hook_quality: "strong" | "adequate" | "weak"
      value_prop_clarity: "clear" | "muddled" | "missing"
      proof_usage: "strong" | "weak" | "none"
      personalization_level: "tier_1" | "tier_2" | "tier_3" | "none"
      cta_quality: "clear_low_friction" | "clear_high_friction" | "unclear" | "missing"
      issues: string[]
      rewrite_suggestions: string[]
    }
  ]

  sequence_architecture: {
    touch_count_verdict: string
    timing_verdict: string
    angle_diversity_verdict: string
    overall_flow: string
  }

  strongest_element: string            # What's working best in the copy
  weakest_element: string              # What needs the most work
}

Step 5: Lead Quality Assessment

Purpose: Evaluate whether we're sending to the right people. Even perfect copy will fail if the audience is wrong. Cross-references reply patterns with lead data.

Input Contract

lead_list_summary: { ... } | null     # From Step 1 (if available)
reply_analysis: { ... }               # From Step 3
quantitative_analysis: { ... }        # From Step 2
icp_definition: string                # From config

Process

A) Targeting Assessment

B) Signal Quality (Inferred from Replies)

C) Responder Profile Analysis

If we have data on who replied positively:

• What titles responded most?
• What industries?
• What company sizes?
• Is there a pattern that differs from the overall send list?

This reveals the "actual ICP" vs. the "intended ICP" — sometimes they diverge.

Output Contract

lead_quality: {
  overall_grade: "A" | "B" | "C" | "D" | "F"
  overall_summary: string

  targeting_assessment: {
    title_fit: "aligned" | "partially_aligned" | "misaligned"
    industry_fit: "aligned" | "partially_aligned" | "misaligned"
    issues: string[]
  }

  signal_quality: {
    primary_signal: string              # What the reply patterns suggest about targeting
    evidence: string[]                  # Specific data points
  }

  responder_profile: {
    most_responsive_titles: string[]
    most_responsive_industries: string[]
    actual_vs_intended_icp: string      # How the responding audience differs from targeting
  } | null

  recommendations: string[]
}

Step 6: Generate Report

Purpose: Synthesize all analyses into a single diagnostic report with two sections: executive summary and detailed breakdown. Plus a prioritized list of actions.

Input Contract

quantitative_analysis: { ... }        # From Step 2
reply_analysis: { ... }               # From Step 3
copy_assessment: { ... }              # From Step 4
lead_quality: { ... }                 # From Step 5
benchmarks: { ... }                   # Calibrated in Step 2

Report Structure

# Sequence Performance Review: [Campaign Name]
**Period:** [date range] | **Status:** [active/paused/completed]

---

## Executive Summary

**Overall verdict:** [One sentence: "This campaign is [performing well / underperforming / in trouble] because [key reason]."]

| Dimension | Grade | One-Line Assessment |
|-----------|-------|-------------------|
| Metrics | [A-F] | [e.g., "Reply rate 2x benchmark — strong performance"] |
| Copy Quality | [A-F] | [e.g., "Touch 1 is strong, Touch 2-3 need new angles"] |
| Lead Quality | [A-F] | [e.g., "Targeting is tight but volume is too low"] |
| Reply Quality | [Strong/Mixed/Weak/Toxic] | [e.g., "60% positive — objections are handleable"] |

### What's Working (Double Down)
- [Specific thing that's working, with data]
- [Another]

### What's Not Working (Fix or Kill)
- [Specific thing that's failing, with data]
- [Another]

### Top 3 Actions
1. [Highest-impact action with expected result]
2. [Second]
3. [Third]

---

## Detailed Metrics

### Overall Performance
| Metric | Actual | Benchmark | Status |
|--------|--------|-----------|--------|
| Emails sent | X | — | — |
| Deliverability | X% | >97% | [flag] |
| Open rate | X% | Y% | [above/below] |
| Reply rate | X% | Y% | [above/below] |
| Positive reply rate | X% | Y% | [above/below] |
| Bounce rate | X% | <3% | [flag] |
| Meeting conversion | X% | — | — |

### Performance by Touch
| Touch | Sent | Open Rate | Reply Rate | Marginal Reply Rate | % of Total Replies |
|-------|------|-----------|------------|--------------------|--------------------|
| 1 | X | Y% | Z% | Z% | W% |
| 2 | X | Y% | Z% | Z% | W% |
| 3 | X | Y% | Z% | Z% | W% |

[Commentary: which touches are carrying the campaign? Is Touch 2+ adding value or wasting sends?]

### Variant Performance (if A/B testing)
| Touch | Variant | Subject | Sent | Open Rate | Reply Rate | Confidence | Action |
|-------|---------|---------|------|-----------|------------|------------|--------|
| 1 | A | "..." | X | Y% | Z% | Likely | Scale |
| 1 | B | "..." | X | Y% | Z% | Likely | Kill |

---

## Reply Deep Dive

### Reply Classification
| Category | Count | % of Replies | Trend |
|----------|-------|-------------|-------|
| Positive interest | X | Y% | — |
| Meeting request | X | Y% | — |
| Warm / Curious | X | Y% | — |
| Objection — Timing | X | Y% | — |
| Objection — Relevance | X | Y% | [flag if high] |
| Not interested | X | Y% | — |
| Auto-reply / OOO | X | Y% | — |

### Top Objections
| Objection | Count | Handleable? | Suggested Response |
|-----------|-------|------------|-------------------|
| [objection] | X | Yes/No | [how to handle or what it implies] |

### Notable Replies
[5-10 most instructive replies with quotes and what they teach us]

---

## Copy Assessment

### Subject Lines
| Touch | Subject | Open Rate | Verdict | Issue |
|-------|---------|-----------|---------|-------|
| 1 | "..." | X% | Working | — |
| 2 | "..." | X% | Underperforming | Too generic |

### Email Body
| Touch | Grade | Hook | Value Prop | Proof | Personalization | CTA |
|-------|-------|------|-----------|-------|----------------|-----|
| 1 | B+ | Strong | Clear | One proof point | Tier 2 | Clear, low-friction |
| 2 | C | Weak ("just following up") | Repeat of T1 | None new | Tier 1 | Same CTA as T1 |

[Specific rewrite suggestions for underperforming touches]

### Sequence Architecture
- Touch count: [verdict]
- Timing: [verdict]
- Angle diversity: [verdict]
- Overall flow: [verdict]

---

## Lead Quality

### Targeting
| Dimension | Intended | Actual (from responders) | Fit |
|-----------|---------|------------------------|-----|
| Titles | VP Sales, CRO | Sales Managers (mostly) | Partial |
| Industries | SaaS, FinTech | SaaS (90%) | Aligned |
| Company size | 50-500 | 20-200 | Shift down |

[Commentary on targeting gaps]

---

## Recommendations (Prioritized)

### High Priority (Do This Week)
1. **[Action]** — [Data point] → [Expected impact]
2. **[Action]** — [Data point] → [Expected impact]

### Medium Priority (Do This Month)
3. **[Action]** — [Data point] → [Expected impact]
4. **[Action]** — [Data point] → [Expected impact]

### Low Priority (Backlog)
5. **[Action]** — [Data point]

### Kill List
- [Anything that should be stopped — a variant, a touch, a segment]

Recommendation Generation Logic

Output Contract

report: {
  executive_summary: string           # Markdown formatted
  detailed_report: string             # Markdown formatted
  recommendations: [
    {
      priority: "high" | "medium" | "low" | "kill"
      category: "subject_line" | "body_copy" | "targeting" | "sequence" | "variant" | "infrastructure" | "list_hygiene"
      action: string
      data_point: string
      expected_impact: string
    }
  ]
  rewrite_suggestions: [              # Ready-to-use copy alternatives
    {
      touch: integer
      element: "subject" | "body" | "cta"
      current: string
      suggested: string
      rationale: string
    }
  ] | null
}

Human Checkpoint

Present the executive summary, then offer drill-down:

[Executive Summary rendered]

---

Full detailed report available with:
- Touch-by-touch breakdown with open/reply rates
- Every reply classified with quotes
- Copy scored on 6 dimensions per touch
- Lead quality assessment with actual vs intended ICP
- Prioritized recommendations with rewrite suggestions

Want to see the full report? Or act on a specific recommendation?

Execution Summary

Total human review time: ~10-15 minutes for an analysis that would take 1-2 hours of manual campaign review.

Adapting to Data Availability

Minimum viable data: Emails sent + reply count + email copy text. Everything else enriches.

Tips

• Run this at Day 7 and Day 14 of a campaign. Day 7 catches early problems (deliverability, subject lines). Day 14 gives enough reply volume for meaningful objection analysis.
• Reply analysis is where the gold is. Metrics tell you WHAT is happening. Replies tell you WHY.
• High open rate + low reply rate is a copy problem. People are interested enough to open but the email doesn't deliver on the subject line's promise.
• Low open rate + decent reply rate (among openers) is a subject line problem. The email works — people just aren't seeing it. Fix the subject line and watch replies scale.
• "Not relevant" objections are the most important signal. If >20% of replies say "this isn't for me," you have a targeting problem, not a copy problem. No amount of copy rewriting fixes a bad list.
• Don't kill a variant too early. You need 100+ sends per variant for directional data. Calling a winner at 30 sends is just reading noise.
• Touch 2/3 should contribute 30-40% of total replies. If Touch 1 accounts for 90%+ of replies, your follow-ups aren't adding value — rewrite them with genuinely new angles.
• Save the objection analysis. It's product marketing gold. The exact language prospects use to object is the language your website and sales deck should preemptively address.