agentleFS
fourteenwmfourteenwm/ppc-ai-skills/portfolio-health-prioritization/SKILL.md

portfolio-health-prioritization

Skill72 starsChanged 4 months agoMITView on GitHub

Account prioritization criteria for portfolio health monitoring and daily briefings. Auto-invoke when determining which accounts need investigation, prioritizing daily actions, running portfolio health checks, or deciding investigation order. Provides 5-tier priority system with portfolio-specific thresholds (Portfolio A ±5%, Portfolio B ±8%).

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Investigation: Not needed unless pattern changes

    line 72
  2. Distinctive:

    No recent issues or alerts

    line 80
  3. Distinctive:

    No shared budgets (need granular control)

    line 104

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “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

Portfolio Health Prioritization Skill

Purpose: Provides standardized criteria for prioritizing Google Ads accounts that need investigation or action in portfolio health monitoring workflows.

Type: Domain knowledge skill (auto-invoked)


What This Skill Deliberately Does NOT Do

  • Investigate accounts. It decides WHICH accounts get investigated and in what order — the digging itself belongs to investigation skills (e.g. underspending-investigation for pacing misses).
  • Run the health check. It consumes the flagged-account list your own monitoring produces (script, sheet, or API sweep — "replace with your own equivalent" throughout) — it ships no scanner.
  • Set the thresholds. The ±5%/±8% tolerances are portfolio policy from portfolio-pacing-rules; this skill applies them to build the day's list, it doesn't choose them.
  • Fix anything. Output is a prioritized list. Budget changes, quality work, and every other intervention happen downstream in their own skills, under their own guardrails.

Quick Reference: Priority Tiers

🚨 Tier 1: CRITICAL (Investigate First)

Criteria:

  • Pacing variance >±15% (any portfolio)
  • Account completely off pace (2x expected or 50% of expected)
  • Major performance degradation (CPA doubled, ROAS halved)

Action Required: Same day, immediately Typical Count: 0-3 accounts per portfolio per day Investigation: Mandatory deep dive


⚠️ Tier 2: HIGH PRIORITY (Investigate Today)

Criteria:

  • Portfolio B/Portfolio C: Pacing variance >±8% (outside tolerance)
  • Portfolio A: Pacing variance >±5% (stricter SLA)
  • Zero spenders (campaigns with $0 spend MTD)
  • Recent pacing deterioration (variance worsening >5% in 7 days)

Action Required: Within 24 hours Typical Count: 3-10 accounts per portfolio per day Investigation: Recommended deep dive


ℹ️ Tier 3: MEDIUM PRIORITY (Investigate This Week)

Criteria:

  • Pacing near threshold (Portfolio B: 5-8%, Portfolio A: 3-5%)
  • Conversion tracking issues (sudden drops)
  • Ad disapprovals (multiple ads affected)
  • Campaign end dates (ending in 7-14 days)

Action Required: Within 7 days Typical Count: 5-15 accounts per portfolio Investigation: Optional, based on capacity


✅ Tier 4: LOW PRIORITY (Monitor)

Criteria:

  • Within tolerance but near threshold
  • Historical patterns (seasonal, expected variance)
  • Stable performance with minor fluctuations

Action Required: Monitor for 3-5 days Typical Count: Most of portfolio Investigation: Not needed unless pattern changes


⭕ Tier 5: NO ACTION NEEDED

Criteria:

  • Pacing within tolerance (Portfolio B: <±8%, Portfolio A: <±5%)
  • Performance acceptable (CPA on goal, ROAS meeting targets)
  • No recent issues or alerts

Action Required: None Typical Count: 70-85% of portfolio Investigation: Not applicable


Portfolio-Specific Thresholds

Portfolio A Portfolio

Pacing Tolerance: ±5% monthly (stricter SLA) Rationale: High-visibility client, frequent reporting, strict budget accountability

Priority Thresholds:

  • 🚨 Critical: >±15%
  • ⚠️ High: >±5%
  • ℹ️ Medium: 3-5%
  • ✅ Low: 2-3%
  • ⭕ On Pace: <±2%

Special Considerations:

  • Brand campaign cap: 15% of total spend (flag if exceeded)
  • ROAS-focused (primary KPI)
  • No shared budgets (need granular control)

Accounts:

  • Acme Plumbing (CID: [CUSTOMER_ID])
  • Best HVAC (CID: [CUSTOMER_ID])
  • City Dental (CID: [CUSTOMER_ID])
  • Account D (multiple accounts)

Portfolio B Portfolio

Pacing Tolerance: ±8% monthly (standard) Rationale: More flexible pacing, focus on cost efficiency

Priority Thresholds:

  • 🚨 Critical: >±15%
  • ⚠️ High: >±8%
  • ℹ️ Medium: 5-8%
  • ✅ Low: 3-5%
  • ⭕ On Pace: <±3%

Special Considerations:

  • CPA-focused (primary KPI, some testing ROAS)
  • Shared budgets common (GEO + Bedroom campaigns)
  • Campaign line filtering (Pmax/Dgen/Search account designations)

Campaign Structure:

  • Original "Core 4": Brand, GEO, Bedroom, GDN Remarketing
  • Current: Added Pmax/Demand Gen, paused GDN Remarketing

Portfolio C Portfolio

Pacing Tolerance: ±8% monthly (same as Portfolio B) Rationale: Mid-tier accounts, similar to Portfolio B management style

Priority Thresholds:

  • 🚨 Critical: >±15%
  • ⚠️ High: >±8%
  • ℹ️ Medium: 5-8%
  • ✅ Low: 3-5%
  • ⭕ On Pace: <±3%

Accounts:

  • Portfolio C - Multi-region (CID: [CUSTOMER_ID])
  • ProClean (CID: [CUSTOMER_ID])
  • Quick Fix (CID: [CUSTOMER_ID])

Prioritization Decision Tree

Step 1: Check Critical Thresholds

START: New account flagged in portfolio health check

Is pacing variance >±15%?
 └─ YES → 🚨 TIER 1 CRITICAL (investigate immediately)
 └─ NO → Continue to Step 2

Step 2: Check Portfolio-Specific Thresholds

Which portfolio is this account in?
 ├─ Portfolio A → Is variance >±5%?
 │ └─ YES → ⚠️ TIER 2 HIGH PRIORITY
 │ └─ NO → Continue to Step 3
 │
 └─ Portfolio B/Portfolio C → Is variance >±8%?
 └─ YES → ⚠️ TIER 2 HIGH PRIORITY
 └─ NO → Continue to Step 3

Step 3: Check Secondary Indicators

Does account have any of these issues?
 ├─ Zero spender (MTD spend = $0) → ⚠️ TIER 2 HIGH PRIORITY
 ├─ Recent deterioration (variance +5% worse in 7 days) → ⚠️ TIER 2 HIGH PRIORITY
 ├─ Conversion tracking issues → ℹ️ TIER 3 MEDIUM PRIORITY
 ├─ Ad disapprovals → ℹ️ TIER 3 MEDIUM PRIORITY
 ├─ Campaign ending soon (7-14 days) → ℹ️ TIER 3 MEDIUM PRIORITY
 └─ None of the above → Continue to Step 4

Step 4: Check Proximity to Threshold

Is variance approaching threshold?
 ├─ Portfolio A: 3-5% → ℹ️ TIER 3 MEDIUM PRIORITY
 ├─ Portfolio B/Portfolio C: 5-8% → ℹ️ TIER 3 MEDIUM PRIORITY
 ├─ Any portfolio: 2-3% → ✅ TIER 4 LOW PRIORITY (monitor)
 └─ Within tolerance → ⭕ TIER 5 NO ACTION NEEDED

Focus Rules (Investigation Order)

Rule 1: Underspending > Overspending

Why:

  • Underspending issues typically have quick fixes (budget increase, remove bid caps)
  • Overspending issues often require strategic changes (pause campaigns, reduce bids, improve quality)

Exception: Overspending >±15% = critical (may exceed client budget)

Practical Application:

Accounts to investigate:
1. Account A: +12% underspending (Portfolio B)
2. Account B: -10% overspending (Portfolio B)
3. Account C: +9% underspending (Portfolio B)

Investigation Order:
1st: Account A (+12% underspending - highest variance, quick fix potential)
2nd: Account C (+9% underspending - quick fix potential)
3rd: Account B (-10% overspending - needs strategic review)

Rule 2: Portfolio A > Other Portfolios

Why:

  • Stricter client SLA (±5% vs ±8%)
  • Higher visibility accounts (larger spend, more frequent reporting)
  • Client escalation risk (Portfolio A has tighter requirements)

Practical Application:

Accounts to investigate:
1. Account A: +6% underspending (Portfolio A)
2. Account B: +9% underspending (Portfolio B)
3. Account C: +6% underspending (Portfolio B)

Investigation Order:
1st: Account A (Portfolio A +6% - outside ±5% tolerance)
2nd: Account B (Portfolio B +9% - higher variance)
3rd: Account C (Portfolio B +6% - within ±8% tolerance, lower priority)

Rule 3: Recent Changes > Historical Patterns

Why:

  • Accounts with recent budget changes may be in ramp-up period (normal, monitor only)
  • Accounts with sudden variance shifts indicate new issues (investigate)
  • Accounts with historical patterns (seasonal) may not need intervention

Practical Application:

Accounts to investigate:
1. Account A: +10% underspending (budget increased 3 days ago)
2. Account B: +10% underspending (was on pace 7 days ago, suddenly shifted)
3. Account C: +10% underspending (historically underspends in Q1 every year)

Investigation Order:
1st: Account B (sudden shift - new issue, investigate root cause)
2nd: Account A (recent budget increase - check if ramp-up on track, may just need monitoring)
3rd: Account C (historical pattern - seasonal, may not need action)

Rule 4: High Variance > Near Threshold

Why:

  • High variance (>±10%) = significant impact, needs immediate attention
  • Near threshold (5-8% Portfolio B, 3-5% Portfolio A) = monitor, may self-correct

Practical Application:

Accounts to investigate:
1. Account A: +12% underspending (Portfolio B)
2. Account B: +6% underspending (Portfolio B)
3. Account C: +5% underspending (Portfolio A)

Investigation Order:
1st: Account A (+12% - high variance, well outside tolerance)
2nd: Account C (+5% Portfolio A - exactly at Portfolio A threshold)
3rd: Account B (+6% Portfolio B - within ±8% tolerance, monitor)

Daily Briefing Workflow Integration

Daily Health Check Use Case

Typical workflow:

  1. Run your portfolio health check (a script that flags issues like pacing variance, zero-spenders, conversion drops, etc. — replace with your own equivalent)
  • Returns 50-100 accounts with various issues
  1. Apply Portfolio Health Prioritization Skill
  • Classify all accounts into 5 tiers
  • Identify top 3-5 accounts for deep investigation
  1. Launch Investigation Agents (parallel)
  • Tier 1 accounts: All investigated (0-3 accounts)
  • Tier 2 accounts: Top 3-5 investigated
  • Tier 3+: Listed in briefing, not investigated
  1. Generate Briefing
  • Critical: {count} accounts
  • High Priority: {count} accounts
  • Investigated: {count} accounts (with findings)
  • Other Issues: {count} accounts (monitoring only)

Selection Algorithm for Daily Prioritization

Goal: Select 3-5 accounts for deep investigation from potentially 20-30 flagged accounts

Algorithm:

# Pseudocode for account selection
accounts = get_all_flagged_accounts()

# Step 1: Separate by tier
tier_1 = [acc for acc in accounts if acc.variance > 15]
tier_2_a = [acc for acc in accounts if acc.portfolio == 'Portfolio A' and acc.variance > 5]
tier_2_other = [acc for acc in accounts if acc.portfolio != 'Portfolio A' and acc.variance > 8]
tier_2_zero = [acc for acc in accounts if acc.mtd_spend == 0]

# Step 2: Prioritize within tiers
tier_2_underspend = [acc for acc in (tier_2_a + tier_2_other) if acc.variance > 0]
tier_2_overspend = [acc for acc in (tier_2_a + tier_2_other) if acc.variance < 0]

# Step 3: Build investigation list (max 5)
investigation_list = []

# All Tier 1 (critical) get investigated
investigation_list.extend(tier_1)

# Fill remaining slots with Tier 2, prioritizing:
# 1. Portfolio A accounts
# 2. Underspending accounts
# 3. Highest variance
remaining_slots = 5 - len(investigation_list)

candidates = (
 sorted(tier_2_a, key=lambda x: abs(x.variance), reverse=True) +
 sorted(tier_2_underspend, key=lambda x: abs(x.variance), reverse=True) +
 sorted(tier_2_overspend, key=lambda x: abs(x.variance), reverse=True) +
 tier_2_zero
)

investigation_list.extend(candidates[:remaining_slots])

return investigation_list[:5] # Max 5 accounts

Real-World Example: Daily Prioritization

Scenario: AI Error Analysis Returns 25 Flagged Accounts

Input Data:

Tier 1 (Critical >±15%): 2 accounts
- Portfolio B - ProClean: +18.5% underspending
- Portfolio A - Best HVAC: -16.2% overspending

Tier 2 (High Priority):
- Portfolio A (>±5%): 3 accounts
 - City Dental: +6.8% underspending
 - Account E: +5.3% underspending
 - Acme Plumbing PCV: -5.1% overspending

- Portfolio B (>±8%): 8 accounts
 - Quick Fix: +12.3% underspending
 - Metro Auto: +9.7% underspending
 - Account G: -10.2% overspending
 - [5 more accounts between ±8-10%]

- Zero Spenders: 3 accounts

Tier 3 (Medium Priority): 9 accounts (near threshold, ad disapprovals, etc.)

Apply Prioritization:

  1. All Tier 1 accounts (2 accounts):
  • Portfolio B - ProClean (+18.5%)
  • Portfolio A - Best HVAC (-16.2%)
  1. Top 3 from Tier 2 (using focus rules):
  • Priority 1: Portfolio B - Quick Fix (+12.3% underspending - high variance + underspending)
  • Priority 2: Portfolio A - City Dental (+6.8% underspending - Portfolio A priority)
  • Priority 3: Portfolio B - Metro Auto (+9.7% underspending - underspending focus)

Investigation List (5 accounts):

  1. Portfolio B - ProClean (+18.5%) - CRITICAL
  2. Portfolio A - Best HVAC (-16.2%) - CRITICAL
  3. Portfolio B - Quick Fix (+12.3%) - HIGH PRIORITY
  4. Portfolio A - City Dental (+6.8%) - HIGH PRIORITY (Portfolio A)
  5. Portfolio B - Metro Auto (+9.7%) - HIGH PRIORITY

Not Investigated (mentioned in briefing only):

  • 5 other Portfolio B accounts with ±8-10% variance
  • 2 other Portfolio A accounts with ±5% variance
  • 3 zero spenders
  • 9 Tier 3 accounts

Briefing Summary:

  • Investigated: 5 accounts (with root cause + recommendations)
  • Other High Priority: 10 accounts (listed, not investigated)
  • Medium Priority: 9 accounts (listed by category)
  • On Pace: ~70 accounts (count only)

When to Use This Skill

Auto-Invoked When:

  • Running daily portfolio health checks
  • Determining which accounts need investigation
  • Prioritizing daily actions across portfolio
  • User asks "what needs my attention today"
  • Portfolio health check workflows
  • Selecting accounts for batch optimizations

Manual Invocation:

  • Custom portfolio queries ("Show me high priority Portfolio B accounts")
  • Weekly/monthly portfolio reviews
  • Client reporting (prioritized account lists)
  • Ad-hoc "what should I look at first?" questions

When to Load Other Skills

SkillWhen
portfolio-pacing-rulesFor the pacing thresholds this skill's tiers are built on (±8% Portfolio B, ±5% Portfolio A) and the budget management philosophy behind them — the tolerance policy lives there
underspending-investigationAfter prioritization, for each selected account whose flag is a pacing miss — it runs the actual investigation
budget-recommendation-calculatorDownstream, when an investigation lands on a budget fix — tier urgency affects how quickly the recommendation moves
account-diagnosticWhen a prioritized account needs a full health inspection rather than a single-issue investigation — deciding which account to inspect first is this skill's job, the inspection is that one's

External pieces you supply yourself: the daily portfolio health check that produces the flagged-account list, and the pacing data source (sheet or API) behind the variance numbers — both referenced throughout this skill, both replaceable with your own equivalents.


Created: 2025-11-01 Based On: Portfolio health monitoring best practices Status: Active

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.