sales-pipeline

Sales Pipeline

Turn pipeline data into a view of what is real, what is at risk, and what should happen next.

Use this skill for

• pipeline health reviews
• weekly forecast calls
• deal inspection and next-step planning
• CRM cleanup and stage hygiene
• coverage and conversion analysis
• diagnosing why opportunities are stalling or slipping

Core model

Analyze pipeline at three levels:

• Pipeline level: coverage, stage mix, aging, conversion, forecast confidence
• Stage level: bottlenecks, drop-off, cycle time, definition drift
• Deal level: champion, pain, timeline, next step, risk, and close realism

Default output structure

1. Pipeline snapshot

• total pipeline
• pipeline by stage
• pipeline by close period
• coverage versus target if available

2. Forecast view

• committed
• best case
• upside
• excluded or unforecastable pipeline

For each bucket, explain why the confidence level is justified.

3. Stage health

• volume by stage
• average age in stage
• stale opportunities
• conversion or exit pattern
• where stage definitions appear weak or inconsistently applied

4. Deal risks

• top deals at risk
• slip risk
• missing next step
• missing buying signal, champion, budget, or decision path

5. Recommended actions

• pipeline cleanup
• deal-level action
• manager inspection point
• process fix

Forecast rules

• Separate pipeline amount from forecast confidence.
• Do not let CRM stage alone determine forecast category.
• Flag any deal with no dated next step as low confidence.
• Flag stale deals explicitly; do not leave them hidden in stage totals.
• Name what evidence would move a deal up or down in confidence.

Default hygiene checks

Run these checks even if the user does not ask for them explicitly:

• missing next meeting or next action
• close date in the past
• opportunity aged beyond normal for its stage
• no recent activity
• stage does not match note history
• no identified champion or decision maker
• no clear reason to buy now

Stage logic

If the user does not supply stage definitions, infer a simple progression such as:

• discovery
• qualification
• evaluation
• proposal
• negotiation
• closed won or lost

Then state that the stage model was inferred.

Failure modes to avoid

• optimistic forecast unsupported by evidence
• summarizing stage totals without inspecting deal quality
• confusing activity with progress
• leaving stale deals inside headline numbers
• recommending actions that are too generic to execute

Quality bar

• Forecast buckets are evidence-based.
• Stale or weak deals are visible.
• Recommended actions map to specific deals, stages, or process problems.
• The user can leave the review knowing what to clean, what to push, and what not to trust.