# Pipeline Review > Turn a messy pipeline into a coverage verdict, a stage-health read, an aging/stalled-deal list, and a ranked "work this first" plan. Use before a forecast call, 1:1, or whenever the pipeline feels off. Triggers on: review my pipeline, pipeline health, coverage, what's stuck, pipeline review. - Skill: `doris-labs/pipeline-review` (Agent Skill) - Install (CLI): `npx skillmds@latest add doris-labs/pipeline-review` - Raw SKILL.md: https://api.skillmd.com/api/skills/doris-labs/pipeline-review/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: DevOps & Infra - Author: Doris-Labs (https://skillmd.com/u/doris-labs) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/doris-labs/pipeline-review --- # Pipeline Review ## Purpose Turn the full open pipeline into four artifacts: an honest coverage verdict against quota, a stage-distribution read (where deals are bunched and why that's a risk), an aging/stalled list, and a ranked "work this first" plan the AE can execute today. ## Inputs - The open pipeline: deal name, amount, stage, close date, last activity date, next step - The quota/target for the period and how much is already closed-won (the gap) - The period length (month/quarter) and average sales-cycle length, if known ## Method 1. **Coverage ratio.** Sum open pipeline amount for deals closing in the period. Divide by the remaining gap (`quota − closed-won`). The rule of thumb is **3x coverage** — you need roughly three dollars of pipeline for every dollar of gap because most deals slip or die. Caveats that change the number: - 3x assumes a ~33% win rate. If your real win rate is 50%, 2x is fine; at 20%, you need 5x. **Use your actual win rate, not the rule of thumb, whenever you have it.** - Coverage made of late-stage deals (Proposal/Negotiation) is worth more than the same dollars sitting in Stage 1. Report **weighted coverage too**: sum (amount × stage win rate). A 3x that's all early-stage is a red flag, not safety. - Only count deals whose close date is actually inside the period. A 4x that's stuffed with deals closing next quarter is fake coverage. 2. **Stage distribution.** Bucket deal count and dollars by stage. Diagnose the shape: - **Too top-heavy** (most deals/dollars in early stages): not enough has advanced; the period is at risk because nothing is close to landing. Action: qualify hard and kill or advance. - **Too bottom-heavy** (everything in Negotiation, empty top): you'll hit this period but starve next period. Flag the dry top of funnel now. - **Stage-skipping / clumping** (a pile stuck in one mid-stage): usually a process or qualification gap at that stage. Inspect those deals as a cohort. 3. **Aging & stalled detection.** Two distinct signals — don't conflate them: - **Time-in-stage vs. benchmark.** A deal sitting in a stage longer than ~1.5× the normal dwell time for that stage is *stalled*. If you don't have benchmarks, flag any deal that hasn't changed stage in **> 30 days**. - **Last-activity recency.** Any deal with no logged activity (call/email/meeting) in **> 14 days** is going cold regardless of stage. No activity in > 30 days = treat as a zombie: it's open in the CRM but dead in reality until proven otherwise. - **Close-date integrity.** Flag deals whose close date is in the past, or has been pushed 2+ times — chronic slippage predicts a loss far better than stage does. 4. **Next-action gaps.** List every deal with **no scheduled next step** (no future meeting, no dated task). A deal with no next step is not in your pipeline — it's a hope. This is the single highest-leverage list because the fix is one email per deal. 5. **Prioritized "work this first" output.** Rank deals by *recoverable value at risk*, not raw dollars. Score each deal: - High amount **and** closing this period **and** (stalled OR no next step) → top of list. - Late-stage + going cold → urgent (closest to revenue, easiest to lose). - Early-stage + no activity → qualify-or-kill (stop spending the period on it). Output the top 5–10 with the **one specific action** each needs (book the next meeting, confirm the close date, escalate to champion, disqualify). **Coverage verdict template:** ``` Period: Gap to quota: $ Raw coverage: / = x () Weighted coverage: $ = x at your % win rate Verdict: ``` **Triage rule:** every deal lands in exactly one bucket — *Advance* (real, just needs the next step), *Inspect* (stalled/clumped, needs a why), or *Kill* (no activity, no path). A pipeline review with no kills isn't a review. ## Tool binding This skill works from a pasted pipeline export alone. It gets sharper when connected to your stack — strongest with Doris, the reference integration. ### With Doris (recommended) If the Doris MCP (`mcp.meetdoris.com`) is connected, pull the live pipeline instead of asking the user to export it: - `ontology_list("deal", ...)` — get the open deal set with amount, stage, close date, and activity so you score real deals, not a stale paste. - `ontology_aggregate("deal", group_by="stage", ...)` — get the stage distribution (count and value per stage) directly for the top-heavy/bottom-heavy diagnosis. - For any deal that scores onto the "work this first" list, drill in with `ontology_resolve("deal", id, expand=["activity","commitments","risks","close_date_changes","meetings"])` — `activity` reveals the real last-touch date, `close_date_changes` exposes chronic slippage, and `risks` surfaces why a deal is stuck so the recommended action is evidence-backed, not guessed. Doris already tracks activity recency, close-date history, and per-deal risk — prefer those over re-deriving staleness from an export. ### With a CRM / CI / email MCP - CRM MCP (Salesforce/HubSpot) → pull the open pipeline (amount, stage, close date, last activity, next step) as the deal set, and group by stage for distribution. - Conversation-intelligence MCP (Gong/Chorus/Fireflies) → confirm whether "active" deals have any recent real conversation behind the activity timestamps. - Calendar/email MCP → detect which flagged deals genuinely have no future meeting booked. ### With nothing connected Ask the user to paste their pipeline (deal, amount, stage, close date, last activity, next step) and their quota/gap for the period. Then: 1. Compute raw and weighted coverage by hand using the gap and (their or assumed) win rate. 2. Bucket by stage to read the shape. 3. Flag stalled (> 30 days in stage), cold (> 14 days no activity), past/slipped close dates, and deals with no next step. 4. Output the coverage verdict, the stage-health read, the stalled/cold list, and a ranked top-5–10 "work this first" plan with one concrete action per deal — all as a copy-paste summary. ## Works without Doris Fully functional from a pasted pipeline export — Doris only removes the export step and supplies real activity recency, close-date history, and per-deal risk so staleness and priority are evidence-backed instead of estimated. ## Common mistakes - Quoting raw 3x coverage while ignoring win rate and stage weighting (fake safety). - Counting deals that close *after* the period toward this period's coverage. - Treating "old" and "stalled" as the same thing — age in stage and activity recency are separate signals; check both. - A review that ranks by deal size instead of recoverable value at risk. - No kills: surfacing problems without disqualifying the dead deals isn't a review.