HG Insights|Phoenix

Portfolio Triage — Monday Screen

Three customers, three verdicts, one Monday-morning ritual.

Portfolio Triage — Monday Screen
Sample preview

Sample output for this workflow will appear here once it is captured.

Run the workflow in Claude, ChatGPT, or Phoenix Playground using the buttons below to see real output.

Overview

Your CSM runs `csm-churn-prediction-time-series` per-account; this is the batch version for the Monday-morning book screen. Pass in 3 customer domains, get a per-account Watch/Yellow/Red verdict with the headline signal for each — rank by verdict severity. Hard cap 3 domains per invocation: the underlying signals are tool-call-heavy and exceeding the cap will exhaust the 25-turn capture budget.

Use cases

  • The Monday ritual that scales to a 30-account book

    Your CSM runs this 10 times (3 accounts each, ≤30 sec per batch), gets a ranked verdict list across the whole book, and walks into the team standup with 'these 4 are Red, here's the headline signal for each, here's the next workflow in the chain'. No more reading 30 dashboards manually.

  • Verdict-anchored next-step

    Every Yellow / Red verdict pairs with a specific next workflow (`csm-adoption-decline-diagnostic` for adoption-driven Reds, `csm-save-play-sequencer` for verdicts that need a save plan). The triage workflow tells you what to run next, instead of dumping you back into the marketplace.

View workflow prompt
# Portfolio Triage — Monday Screen

## Parameters

- `{{domains}}` *(required)* — Comma-separated customer domains. HARD CAP 3 per invocation; re-run in batches for larger books.. Example: `customer1.com,customer2.com,customer3.com`
- `{{your_product}}` *(required)* — Your product or category — anchors per-domain churn screens. Example: `HG Insights`

## Purpose
Batch version of `csm-churn-prediction-time-series` for the Monday-morning book screen. Hard cap 3 domains per invocation — the underlying signals are tool-call-heavy and exceeding the cap will exhaust the 25-turn capture budget. Re-run in batches of 3 for larger books.

## Process
For EACH domain in {{domains}} (max 3):
1. **Adoption** — `company_install_time_series` filtered to {{your_product}}, `timeRange: last_6_months`. Capture `intensity_momentum` + % drop in `current_intensity`.
2. **Competitor scan** — `company_technographic` filtered to {{your_product}}'s category. Flag any vendor with `firstVerifiedDate` in last 90 days.
3. **Intent decay** — `list_intent_topics` (once for the category, shared across all domains) → `company_intent` per domain with `limit: 200`. Post-fetch filter `topics[]` for the resolved topic_ids; check `trend`.
4. **Per-domain verdict** — apply the `csm-churn-signal-weighting` skill rubric (lightweight: adoption decline OR competitor add OR intent decay → Yellow; two of those → Red).
5. **Rank** — order domains by verdict severity (Red > Yellow > Watch); within same verdict, by signal count.

## Output Format
- `# 📊 Monday Portfolio Screen — Watch List`
- `## Ranked verdicts` — table: rank | domain | verdict | headline signal (one sentence)
- `## Actions this week` — for each Yellow/Red, one bullet naming the next step (e.g., "Run `csm-adoption-decline-diagnostic` on customer1.com" — the next workflow in the chain)

## Quality Checklist
- Hard cap 3 domains enforced — if {{domains}} contains more, process the first 3 and instruct user to re-run
- Each verdict cites its driving signal
- Tools shared across domains where possible (e.g., `list_intent_topics` once for the category)
- Cap total tool calls at ~12 across the 3 domains