Prioritize Initiatives (ICE Scoring)

Solution Track — Step 2 of 4. Ranks initiatives with evidence-backed ICE scoring.

Inputs Required

Initiative list from .agents/mkt/initiatives.md

Output

.agents/mkt/prioritized-initiatives.md

Quality Gate

Before delivering, verify:

Every score (I, C, E) has a one-sentence evidence citation — no naked numbers
Scores are differentiated: no more than 2 initiatives share the same total ICE score
≤3 initiatives above cut line (force constraint even if team wants more)
Each "proceed" initiative has: named owner, target metric with baseline, kill criteria

Chain Position

Previous: mkt-initiative-generate | Next: mkt-funnel-planner

Before Starting

Step 0: Product Context

Check for .agents/mkt/product-context.md. If missing: INTERVIEW. Ask the user 8 product questions (what, who, problem, differentiator, proof points, pricing, objections, voice) and save to .agents/mkt/product-context.md. Or recommend running mkt-copywriting to bootstrap it.

Required Artifacts

Artifact	Source	If Missing
`initiatives.md`	mkt-initiative-generate	STOP. "Run mkt-initiative-generate first to create the initiative list."

Optional Artifacts

Artifact	Source	Benefit
`targets.md`	mkt-funnel-planner	Scoring calibration

Read .agents/mkt/initiatives.md. List the initiatives.

Step 1: Rank BEFORE Scoring

Critical: Rank all initiatives from most to least promising FIRST, based on gut + evidence. Then assign scores. This prevents the "everything is a 6" clustering problem.

Ask the user: "If you could only do ONE of these, which would it be? And after that one?"

Force-rank all initiatives 1 through N before scoring.

Step 2: Score Each Initiative

Score 1-10 on three dimensions. Every score needs a one-sentence evidence citation.

Dimension	1-3	4-6	7-10
Impact — How much will it move the target metric?	Minor, barely noticeable	Meaningful improvement	Significant or transformative
Confidence — How sure are we?	Gut feeling, no data	Some supporting data	Direct evidence, proven playbook
Ease — How easy to execute?	Multi-team, months	Moderate effort, manageable	Small team, quick turnaround

Bad: "Impact: 7" Good: "Impact: 7 — paid traffic is 60% of signups, and the old targeting had 3.5% conversion vs. current 1.2%"

WebSearch directive: If the user lacks evidence for a score, search for supporting data: "[tactic] conversion improvement case study", "[similar company] [tactic] results", "[metric] improvement benchmark"

ICE Score = I + C + E (sum). Scores should reflect the forced ranking from Step 1 — if you ranked it #1, its ICE should be highest.

Step 3: Cut Line

Ask: "How many can you execute well simultaneously?"

Push for honesty:

Solo/2-person team: 1-2
Small team (3-5): 2-3
Growth team (5+): 3-4

For each initiative, assign a decision:

Decision	Meaning
Proceed	Start this week. Has owner, metric, kill criteria.
Park	Promising, but can't resource now. Revisit after current batch.
Kill	Not worth doing. State why clearly.

Step 4: Validate "Proceed" Initiatives

For each "Proceed" initiative, confirm:

Owner named (a person, not "the team")
Target metric has a numeric baseline
Kill criteria: "Stop if [metric] < [threshold] after [duration]"
Can begin execution within 1 week

If any check fails, resolve it now — don't proceed with gaps.

Artifact Template

On re-run: rename existing artifact to prioritized-initiatives.v[N].md and create new with incremented version.

---
skill: mkt-initiative-prioritize
version: 1
date: {{today}}
status: draft
---

# Prioritized Initiatives

## Forced Ranking (gut + evidence)

1. [Name] — most promising because [reason]
2. [Name]
3. [Name]
...

## ICE Scoring

| Rank | Initiative | I | C | E | ICE | Key Evidence |
|------|-----------|---|---|---|-----|-------------|
| 1 | [Name] | [score] | [score] | [score] | [sum] | [One-sentence evidence for top-scoring dimension] |
| 2 | ... | ... | ... | ... | ... | ... |

## Decisions

| Initiative | Decision | Owner | Target Metric (Baseline) | Kill Criteria |
|------------|----------|-------|-------------------------|---------------|
| [Name] | Proceed | [Person] | [Metric]: [current] | Stop if [metric] < [threshold] after [duration] |
| [Name] | Park | — | — | — |
| [Name] | Kill | — | — | Reason: [why] |

**Cut line:** [N] initiatives proceeding. Team capacity: [context].

## Next Step

Run `mkt-funnel-planner` to set targets for each "Proceed" initiative.

Worked Example

# Prioritized Initiatives

**Date:** 2026-03-13
**Skill:** mkt-initiative-prioritize

## Forced Ranking

1. Restore + Refine Paid Targeting — fastest fix for biggest root cause
2. Restore Homepage Social Proof — fast, addresses second root cause
3. Homepage A/B: Old vs New — definitive answer on redesign impact
4. Landing Page per Ad Group — higher effort but compounds with #1
5. Intent-Qualifying Ad Creative — interesting but less certain
6. Dedicated Onboarding for Paid — too complex for now

## ICE Scoring

| Rank | Initiative | I | C | E | ICE | Key Evidence |
|------|-----------|---|---|---|-----|-------------|
| 1 | Restore Paid Targeting | 8 | 8 | 9 | 25 | I:8 — paid is 60% of traffic, old targeting converted at 3.5% vs 1.2% now |
| 2 | Restore Social Proof | 7 | 7 | 9 | 23 | C:7 — Wayback shows old page had logos+testimonial that new page removed |
| 3 | Homepage A/B | 6 | 9 | 7 | 22 | C:9 — direct comparison eliminates guessing about redesign impact |
| 4 | LP per Ad Group | 7 | 6 | 5 | 18 | I:7 — message match lifts are well-documented (CXL: 20-50% improvement) |
| 5 | Intent Ad Creative | 5 | 4 | 6 | 15 | C:4 — no precedent with our audience, hypothesis only |
| 6 | Paid Onboarding | 7 | 3 | 3 | 13 | E:3 — needs eng + dynamic content, 4+ weeks |

## Decisions

| Initiative | Decision | Owner | Target Metric (Baseline) | Kill Criteria |
|------------|----------|-------|-------------------------|---------------|
| Restore Paid Targeting | **Proceed** | Sarah | Paid conversion: 1.2% | Stop if <2% after 10 days |
| Restore Social Proof | **Proceed** | James | Homepage bounce: 52% | Stop if >48% after 7 days |
| Homepage A/B | **Park** | — | — | Do after #1 and #2 ship |
| LP per Ad Group | **Park** | — | — | Revisit if paid targeting fix isn't enough |
| Intent Ad Creative | **Kill** | — | — | Too uncertain; paid targeting fix addresses the same root cause more directly |
| Paid Onboarding | **Kill** | — | — | Too complex for current team capacity |

**Cut line:** 2 initiatives proceeding. Team: 3-person growth team.

## Next Step

Run `mkt-funnel-planner` to set targets for Restore Paid Targeting and Restore Social Proof.

References

references/ice-scoring-rubric.md — Detailed scoring calibration with scored examples