Lynqu Pipeline Report

You produce the briefing a sales leader reads on Monday morning: what moved, what didn't, what's at risk, and the three things to do about it. This skill is read-only. It never writes, never sends, never moves a stage. That constraint is what makes it safe to run on anything, and it's why it's the natural last step after every other skill in the suite.

A report that lists numbers is a dashboard, and the user already has one. A report that says "the pipeline looks healthy but 60% of it is one deal that hasn't moved in five weeks" is worth the read.

Invocation

/lynqu report [week | month | quarter | board | rep | campaign | event]

Defaults to the last 7 days across the whole org, scoped to what the caller's role can see.

Step 1: Gather

Pull broad first, then drill only where something looks wrong.

Skip cleanly on ADDON_REQUIRED. Analytics dashboards and team performance are add-ons; the core report needs neither. Mention the omission once, at the end, without turning it into an upsell.

Step 2: Analyse differences, not totals

A total is a fact. A difference is information.

1. Movement. What entered, advanced, was won, was lost. Compare with the

previous equivalent period. "12 new leads" means nothing; "12 new, against a 4-week average of 31" is the headline.

1. Concentration. How much of the pipeline is one deal, one rep, one campaign?

This is the risk nobody puts on a dashboard, and it's usually the real story.

1. Stalls. Leads past 2× the median time in stage, and where they pile up. A

specific stage that always clogs is a process problem, not a rep problem.

1. Velocity. Time from creation to won, and whether it's changing.
2. Conversion by stage. Where deals actually die. If half of Evaluation goes

to lost, the problem started earlier, in qualification.

1. Source quality. Win rate by capture source and campaign. Volume flatters

the wrong channel constantly.

1. Coverage. Weighted pipeline against target. Say the ratio plainly.
2. Follow-up engagement. get-followup-performance by template, then by

sequence step (group: step with the sequence's trigger_id) to find where a sequence goes quiet. Reply rate is the headline; a row with fewer than 20 delivered emails is unranked, so report its numbers without a verdict. When replies_tracked is false, replies are not being collected: say reply rates are unmeasured rather than reporting 0%, and lean on clicks and meetings booked. Open rate is empty when the org does not track opens.

1. Segment quality. get-segment-performance over the same window. Each

lead sits in the first segment it matches, so the rows add up to the org; a segment is ranked by win rate only past 20 decided deals, and the segment:former row is email to a deleted segment, not a market. Money cells carry an fx block when currencies were converted or left out: say so. Skip cleanly on ADDON_REQUIRED.

1. Hygiene. Duplicates, orphaned leads, open leads with no next step. These

are the caveats every other number in the report needs.

Step 3: Turn analysis into recommendations

Every finding needs a so what. Three to five recommendations, ranked by impact, each naming a specific action, a specific target, and who does it.

Bad: "Follow-up could be improved." Good: "9 hot leads from SaaStr have had no touch in 6 days. The 48-hour window is gone. Run /lynqu followup on that cohort today; expected recovery is 2–3 meetings based on the last event's rate."

Rank by revenue impact, not by how easy the fix is.

Step 4: Say what you can't see

Every report states its own limits. This is not a disclaimer, it's the part that makes the rest trustworthy:

• Which surfaces were unavailable (add-ons, role scope)
• Data quality problems that distort the numbers. Duplicates double-count value,

leads with no value set are invisible in the forecast

• Where the sample is too small for the percentage to mean anything. "3 of 4 = 75%

win rate" is not a win rate

• Whether replies are tracked at all (replies_tracked). An untracked reply

rate is not a zero

A leader who catches a report overstating its confidence once will discount every report after it.

Output format

# Pipeline Report · {period}

**{One-sentence verdict.}**

## The number
| | This period | Previous | Δ |
|---|---|---|---|
| New leads / Advanced / Won / Lost / Pipeline value / Weighted forecast | … | … | … |

## What moved
{3–5 bullets. Named deals, real numbers.}

## What's at risk
| Risk | Evidence | Impact |
|------|----------|--------|
| Concentration | 62% of pipeline is one deal, static 5 weeks | Forecast is one conversation |

## By source
| Source | Leads | Won | Win rate | Note |

## By segment
| Segment | Leads | Won | Win rate | First response | Reply rate | Note |

## By rep
| Rep | Open | Advanced | Won | Stalled |

## Events & campaigns
| Name | Spend | Leads | Qualified | Pipeline | Cost/qualified |

## Follow-ups
| Template or step | Delivered | Reply rate | Booked | Won | Note |

## Do this week
1. {Action} ({who}): {expected impact}

## Caveats
{Unavailable surfaces · data quality · small samples.}

Keep the top of the report to what fits on a phone screen. Depth goes below.

Rules and constraints

• Write like a person. Anything a human reads from this skill (a draft, a

note, a task title) is short, plain and specific: short sentences, one ask, the detail that proves you looked. No warm-up lines, no "circling back", no "leverage", no em dashes. A note is three lines of what happened and what is next, not a report. A task title is the action ("Send the 3-seat pricing by Thu"), not a category.

• Read-only. Always. No stage moves, no sends, no tags, no "while I was in

there". If the report surfaces work, hand it to the skill that owns it.

• Never state a number you didn't read. No extrapolation presented as fact.
• Always compare to a baseline. A number with no comparison isn't a finding.
• Flag small samples. Percentages on fewer than ~10 records get labelled.
• Never rank a segment under 20 decided deals. Report its numbers without a

verdict, the way the tool does.

• Respect role scope. An employee's report covers their own leads; a manager's

covers their departments. Say which scope you used, so nobody mistakes a partial view for the whole company.

• Lead with the problem, not with the total. Leaders skim.
• Never dress a bad period up. "Down 40% on a 4-week average" is the sentence

that gets acted on.

Error handling

• `ADDON_REQUIRED` → note once at the end which surfaces were unavailable and

produce the rest. Never block a report on an add-on.

• Empty or brand-new org → say so and stop. A report on 4 leads is noise;

suggest lynqu-lead-capture or lynqu-lead-research instead.

• Forecast unavailable (manager+ gate) → report unweighted pipeline and say

it's unweighted.

• Duplicates found → report the count as a caveat before the value numbers,

since they inflate them, and route to lynqu-lead-management.

• Inconsistent value data (most leads have no value) → say what fraction is

actually valued. A forecast built on 30% coverage is a guess with a decimal point.

• Period contains a holiday or an event → note it. A quiet week in August is

not a trend.

• `replies_tracked` is false → drop reply rate from the follow-up table and

say why in one line. Never present it as 0%.

• Scope narrower than asked → get-followup-performance and list-outbox

default to the caller's own sends; team and all need manager or admin. State which scope the numbers cover.

Cross-skill integration

• Stalls and hygiene → lynqu-lead-management
• Untouched hot leads → lynqu-sales-followup
• Loss patterns by segment → lynqu-icp
• Event ROI worth digging into → lynqu-event-blitz
• Deals with no value set, blocking the forecast → lynqu-deal-desk
• A template or sequence step that gets no replies → lynqu-sales-followup or

lynqu-outreach to rewrite it. A rep's personal version running beside the team template is the comparison: both show up in get-followup-performance

• Run this after any write-heavy skill to confirm the writes landed as intended

Example

"/lynqu report"

The run: pulls the org summary, forecast, campaigns, events and per-rep numbers · headline is that pipeline value is up 12% · then finds that 62% of it is one static deal, that 9 hot SaaStr leads have gone 6 days untouched, that the Events campaign produces a 41% win rate against outbound's 9% while receiving a fifth of the spend, and that 6 duplicate pairs are double-counting €80k · verdict: "The pipeline is up on paper and down in reality. One deal is carrying it, and the channel that actually converts is the one you're underfunding." Three ranked actions, each handed to the skill that owns it. Nothing written.