agentleFS
automatewithudayautomatewithuday/gtm-prospecting-system/skills/pipeline-review/SKILL.md

pipeline-review

Skill1 starsChanged 59 days agoMITView on GitHub

Use when stage 20 of the prospecting system needs the weekly review — where the workflow is leaking leads between stages, which segments and sources are producing positive replies, and what to change in targeting next week. Also use standalone to diagnose outbound campaign performance.

13 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 13 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run uv run scripts/check_run.py runs/<run> — it prints the stage funnel (rows in → rows out) and flags the biggest drop.

    line 17
  2. Distinctive:

    Every reply should be classified:

    line 152
  3. Distinctive:

    Always compare apples to apples.

    line 293
  4. Distinctive:

    Don't average across campaigns.

    line 296

“Common” is measured: kept by at least 6 in 10 of the 13 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Campaign Performance Analysis

Pipeline contract

Stage 20 of 20. Reads the whole run folder. Writes 20-review-<YYYY-MM-DD>.md.

  1. Run uv run scripts/check_run.py runs/<run> — it prints the stage funnel (rows in → rows out) and flags the biggest drop. Start the review there.
  2. Cut positive reply rate and meetings by segment, source_type, and tier. Below 30 sends in a cell, report the count and draw no conclusion.
  3. Propose targeting changes (ICP filters, judge prompt, score weights, sources to add or drop) as a list for the human. Do not edit 02-client-profile.yaml, the judge prompt, or weights yourself.
  4. Approved changes start the next run at stage 2.

Weekly operating cadence: references/cold-email-weekly-rhythm.md. Benchmarks: references/campaign-benchmarking.md.

Files live in runs/<run>/. Column definitions: /prospecting-system. Standalone use: ignore this block.

When to Use

  • Reviewing campaign results after a send cycle completes
  • Diagnosing why a campaign is underperforming
  • Comparing campaign variants to identify winners
  • Building performance reports for clients or leadership
  • Making data-driven decisions about what to scale, pause, or kill

Framework

The Campaign Metrics Stack

Measure campaigns in layers. Each layer tells you something different about what's working and what isn't.

Layer 1: Deliverability     → Did the email reach the inbox?
Layer 2: Engagement         → Did they open and read it?
Layer 3: Response           → Did they reply?
Layer 4: Quality            → Was the reply positive?
Layer 5: Conversion         → Did the reply turn into a meeting?
Layer 6: Revenue            → Did the meeting turn into revenue?

Critical insight: Most teams obsess over Layer 2-3 (opens and replies) while ignoring Layer 1 (deliverability) and Layer 4-6 (quality and revenue). The full stack matters.


Layer 1: Deliverability Metrics

MetricTargetRed FlagWhat It Tells You
Bounce rate< 3%> 5%List quality. High bounces = bad data or unverified emails.
Spam complaint rate0%> 0.1%Content or targeting issue. Immediate pause needed.
Inbox placement rate> 95%< 85%Domain/IP reputation. May need warmup or domain rotation.
Unsubscribe rate< 0.5%> 1%Targeting accuracy. Wrong people or too many touches.

If deliverability metrics are off, STOP. Do not optimize copy, subject lines, or targeting until deliverability is fixed. You're optimizing something that never reaches the prospect.

Deliverability Diagnostic

Bounce rate > 5%?
├── YES → Verify email list. Switch to verified-only sends.
└── NO ↓

Inbox placement < 85%?
├── YES → Check domain reputation (Google Postmaster, MXToolbox)
│         → Reduce daily send volume
│         → Rotate sending domains
│         → Review email content for spam triggers
└── NO ↓

Spam complaints > 0?
├── YES → Review targeting. Are you emailing the right people?
│         → Check for spam trigger words in subject/body
│         → Ensure unsubscribe link is visible
└── NO → Deliverability is healthy. Move to Layer 2.

Layer 2: Engagement Metrics

MetricTargetRed FlagWhat It Tells You
Open rate60-80%< 40%Subject line + deliverability + send timing
Open rate by stepDeclining 5-10% per stepDrops > 20% per stepSequence fatigue or deliverability degradation
Unique opensTrack vs. total opensLow unique/high totalSame people reopening, not new engagement

Note on open tracking: Open tracking uses invisible pixels that can hurt deliverability. If you're tracking opens, use it for diagnostic purposes but consider turning it off for production campaigns. Low open rate with high reply rate = your emails are landing but the pixel isn't loading (common with Outlook).

Subject Line Analysis

Subject LineOpen RateSample SizeVerdict
Variant A__%__ sendsWinner / Loser / Inconclusive
Variant B__%__ sendsWinner / Loser / Inconclusive
Variant C__%__ sendsWinner / Loser / Inconclusive

Minimum sample size: 200 sends per variant before drawing conclusions. Below 200, the variance is too high to trust.


Layer 3: Response Metrics

MetricTargetRed FlagWhat It Tells You
Total reply rate8-15%< 5%Copy relevance + offer strength + list quality
Reply rate by stepStep 1: highest, decliningStep 1 < 3%Opening email isn't earning replies
Reply timingMost replies within 24hDelayed replies (3+ days)Email is interesting but not urgent
Auto-reply rateTrack separately> 10%Many OOO or bounced-to-assistant

Reply Rate Diagnostic

Reply rate < 5% overall?
├── Open rate > 60%?
│   ├── YES → Copy problem. Emails are being read but not compelling.
│   │         → Test new body copy variants
│   │         → Test new CTAs
│   │         → Review personalization quality
│   └── NO  → Subject line or deliverability problem.
│             → Fix opens first, then reassess replies.
│
Reply rate > 5% but meetings are low?
├── Most replies are negative ("not interested", "remove me")?
│   ├── YES → Targeting problem. You're reaching the wrong people.
│   │         → Review ICP and list quality
│   └── NO  → Qualification problem. Interested leads aren't converting.
│             → Review your meeting booking process
│             → Improve reply handling speed

Layer 4: Quality Metrics (Most Important)

MetricTargetRed FlagWhat It Tells You
Positive reply rate2-8%< 1%True message-market fit
Negative reply rate< 3%> 5%Targeting accuracy
Referral repliesTrack separately—Valuable — contact the referral
"Not now" repliesTrack separately—Future pipeline. Add to nurture.

Reply Classification Framework

Every reply should be classified:

CategoryDefinitionActionExample
Positive: MeetingAgreed to a call/meetingBook immediately (< 1 hour)"Sure, let's chat next week"
Positive: InterestInterested but no commitment yetReply with value, ask for meeting"Tell me more about this"
Positive: ReferralDirected you to someone elseContact the referral, thank the original"Talk to Sarah, she handles this"
Neutral: Not NowTiming issue, not a rejectionAdd to nurture, follow up in 30-60 days"Not a priority right now"
Neutral: OOOOut of office auto-replyRe-send after they return"I'm out until March 1"
Negative: Not InterestedPolite rejectionRespect it. Remove from sequence."Thanks, but we're all set"
Negative: AngryHostile responseRemove immediately. Review targeting."Stop emailing me"
Negative: UnsubscribeRequests removalRemove immediately. Legal requirement."Please remove me from your list"

The ratio that matters most:

Positive Reply Rate = (Positive replies ÷ Total emails sent) × 100

This single metric is the truest measure of campaign effectiveness. It strips out noise (opens, negative replies) and measures real demand generation.


Layer 5: Conversion Metrics

MetricTargetRed FlagWhat It Tells You
Reply-to-meeting rate40-60%< 25%Reply handling quality + qualification
Meeting show rate80-90%< 70%Confirmation process + prospect quality
Meeting-to-opportunity rate30-50%< 20%Discovery call quality + true fit
Cost per meetingVaries by ACVRising over timeCampaign efficiency trending

Layer 6: Revenue Metrics

MetricTargetWhat It Tells You
Pipeline generatedTrack per campaignTotal $ value of opportunities created
Revenue closedTrack per campaignActual revenue attributable to outbound
Cost per opportunityVaries by ACVEfficiency of the outbound motion
CAC (Customer Acquisition Cost)< 1/3 of first-year ACVSustainability of the channel
Time to closeTrack by campaign/segmentWhich campaigns produce faster deals

The Performance Analysis Report

Campaign Summary Card

CAMPAIGN: {{campaign_name}}
Period: {{start_date}} — {{end_date}}
Total sends: {{total_sends}}

DELIVERABILITY
  Bounce rate: __% (target: <3%)
  Spam complaints: __ (target: 0)

ENGAGEMENT
  Open rate: __% (target: 60-80%)

RESPONSE
  Total reply rate: __% (target: 8-15%)
  Positive reply rate: __% (target: 2-8%)
  Negative reply rate: __%

CONVERSION
  Meetings booked: __
  Reply-to-meeting rate: __%
  Cost per meeting: $__

REVENUE
  Pipeline generated: $__
  Revenue closed: $__
  ROI: __x

Variant Comparison Table

| Metric | Variant A | Variant B | Variant C | Winner |
|--------|-----------|-----------|-----------|--------|
| Sends | __ | __ | __ | — |
| Open rate | __% | __% | __% | __ |
| Reply rate | __% | __% | __% | __ |
| Positive reply rate | __% | __% | __% | __ |
| Meetings | __ | __ | __ | __ |

Step Performance Breakdown

| Step | Sends | Opens | Open Rate | Replies | Reply Rate | Positive | Positive Rate |
|------|-------|-------|-----------|---------|------------|----------|---------------|
| 1 | __ | __ | __% | __ | __% | __ | __% |
| 2 | __ | __ | __% | __ | __% | __ | __% |
| 3 | __ | __ | __% | __ | __% | __ | __% |
| 4 | __ | __ | __% | __ | __% | __ | __% |

Decision Framework: Scale, Optimize, or Kill

ScenarioDiagnosisAction
Positive reply rate > 5%Campaign is workingScale: increase volume, clone for new segments
Positive reply rate 2-5%Campaign has potentialOptimize: test new variants, improve personalization
Positive reply rate 1-2%Campaign is marginalDiagnose: is it copy, targeting, or timing? Fix one variable
Positive reply rate < 1%Campaign isn't workingKill: stop sending. Redesign from scratch or abandon the angle

Templates

Weekly Campaign Report Template

# Weekly Campaign Report
# Period: {{week}}

## Summary
- Active campaigns: __
- Total sends this week: __
- Avg positive reply rate: __%
- Meetings booked: __

## Top Performing Campaign
{{campaign_name}} — {{positive_reply_rate}}% positive reply rate

## Underperforming (Action Needed)
{{campaign_name}} — {{issue}} → {{recommended_action}}

## Key Decisions
- Scale: {{campaign_to_scale}}
- Optimize: {{campaign_to_optimize}}
- Kill: {{campaign_to_kill}}

## Next Week Plan
- {{action_1}}
- {{action_2}}

Tips

  • Positive reply rate is the only metric that matters for campaign-level decisions. Everything else (opens, total replies) is diagnostic — useful for understanding WHY, but not for deciding WHAT to do.
  • Always compare apples to apples. A campaign targeting VPs at enterprise companies will have different benchmarks than one targeting founders at SMBs. Segment your analysis.
  • The most common analysis mistake: drawing conclusions from too-small samples. 50 sends with a 10% reply rate is not a winning campaign — it's noise. Wait for 200+ sends per variant.
  • Track metrics over time, not just snapshots. A campaign with a 5% reply rate in week 1 and 2% in week 4 is degrading — the list might be exhausted or deliverability is dropping.
  • Don't average across campaigns. One campaign with a 15% positive reply rate and one with 0.5% averages to 7.75% — but that hides the fact that one is amazing and one is terrible.

Progressive disclosure: load platform-specific analytics integrations and automated reporting configs only when analyzing data from a specific sending platform.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.