priority-matrix

name	priority-matrix
description	Final ranking step. Scores each pocket on volume × ACV × win-rate signal, recommends campaign archetype per pocket, and surfaces asymmetric bets (low frequency, high win rate). Use after pocket packaging is complete.

Priority matrix

Pockets aren't equal. Two might both look like real campaigns, but one is your top revenue lever and the other is a quiet asymmetric bet that wants a different treatment.

Inputs

• ./icp-output/pockets/*.md
• ./icp-output/triggers/distribution.md (for frequency and win-rate)
• CRM data via stack.crm if available (for ACV)

Scoring

Each pocket gets three scores:

• Volume (1–5): number of calls in the last 90 days mapping to this pocket. Bucketed: 1 = 1–5 calls, 2 = 6–15, 3 = 16–35, 4 = 36–80, 5 = 81+.
• ACV signal (1–5): average closed-won deal size for accounts in this pocket. If CRM unavailable, use a proxy from the user (or skip and note).
• Win rate signal (1–5): % of stage-qualified calls in this pocket that closed-won. Bucketed: 1 = <10%, 2 = 10–20%, 3 = 20–35%, 4 = 35–50%, 5 = 50%+. If CRM unavailable, use a proxy from outcome tags in the per-call files.

Composite score: simple sum (3–15). Note: composite alone is misleading. The shape of the scores matters.

Pocket shapes (the article's insight)

The article calls out two patterns explicitly:

• Top pocket = highest volume × clear ROI → recommend mass-outbound. Treat as the #1 campaign.
• Lowest-frequency pocket can have the highest win rate → asymmetric bet. Recommend light-touch-asymmetric (high-quality outreach, no mass campaign).

Other shapes:

• High volume, low ACV, low win rate → de-prioritize. Likely a leak.
• Low volume, high ACV, medium win rate → ABM. One-by-one.
• Medium everything but trigger has no current campaign → fastest win. The team has working evidence and zero existing motion to compete with.

Output

Write ./icp-output/priority-matrix.md:

# Priority matrix

| Rank | Pocket | Volume | ACV | Win rate | Composite | Archetype | Note |
|---|---|---|---|---|---|---|---|
| 1 | <Pocket Name> | 5 | 4 | 4 | 13 | mass-outbound | top revenue pocket |
| 2 | <Pocket Name> | 2 | 5 | 5 | 12 | light-touch-asymmetric | high win rate, low frequency — don't mass-mail |
| 3 | … | | | | | | |

## Headline asymmetric bets

Pockets with win rate ≥ 4 but volume ≤ 2:
- <Pocket> — recommend a 50–100 account ABM list run by AEs, not BDRs.

## Pockets with no current campaign

From `triggers/distribution.md`'s coverage gap:
- <Pocket> — campaign should be built. Estimated lift: <reasoning>.

## Pockets to de-prioritize

Composite ≤ 6 or volume + ACV both ≤ 2:
- <Pocket> — call out specifically why and what to do instead (e.g., "convert to inbound nurture only").

## Surprise check

The article notes the original primary persona at Attention turned out to be third by volume. Check yours:

- Which pocket did the team **think** was #1 before this exercise?
- Where did it actually land?
- If it's not #1, name that gap explicitly. The whole point of running the exercise is to surface this kind of misalignment.

Anti-ICP

After the priority matrix, write ./icp-output/anti-icp.md:

# Anti-ICP — who NOT to pursue

Distilled from the per-call files of stalled / no-decision / lost calls where the loss was structural (not execution):

- <attribute pattern> — <evidence: stalled or lost calls with this pattern>
- <attribute pattern> — …

State update

{
  "step": "7",
  "step_status": "completed",
  "ranked_pocket_count": <n>,
  "asymmetric_bets": ["..."],
  "deprioritized": ["..."]
}