Skip to main content

How does value flow — and where does time die?

Flow Efficiency: 0.7% — 80min work in 13 days. World-class is >40%.

Full ASCII diagram
VALUE STREAM MAP: SALES DEV AGENT
════════════════════════════════════════════════════════════

[TRIGGER: Pipeline empty, 0 scored prospects this week]


┌──────────────────┐
│ 1. FIND PROSPECT │ C/T: 30min Wait: 0 %C&A: 40%
│ LinkedIn browsing │
│ Manual search │
└────────┬──────────┘
│ ░░░░░░░░░░░░░░░░░░░░░ (2-3 days: no system, gets deprioritised)

┌──────────────────┐
│ 2. ASSESS FIT │ C/T: 15min Wait: 2d %C&A: 30%
│ Gut feeling │
│ No scoring model │
└────────┬──────────┘
│ ░░░ (1 hour: context switching)

┌──────────────────┐
│ 3. WRITE MESSAGE │ C/T: 20min Wait: 1h %C&A: 15%
│ From scratch │
│ Same template │
└────────┬──────────┘
│ ░░░░░░░░░░░░░ (1-2 days: "I'll send it tomorrow")

┌──────────────────┐
│ 4. SEND + LOG │ C/T: 5min Wait: 1.5d %C&A: 50%
│ Manual send │
│ Forget to log │
└────────┬──────────┘
│ ░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░ (5-14 days: follow-up forgotten)

┌──────────────────┐
│ 5. FOLLOW UP │ C/T: 10min Wait: 7d %C&A: 20%
│ If remembered │
│ Same message │
└────────┬──────────┘


[OUTPUT: Maybe a reply. No data on what worked. No scored pipeline.]

NOTE: %C&A values are estimates (no historical data). First real
measurement comes from Track 3 dogfood outreach.

Cycle Time: 80 min (1.3 hours of actual work)
Wait Time: ~12 days (research gaps + procrastination + forgotten follow-up)
Lead Time: ~13 days
Flow Efficiency: 80min / 13 days = ~0.7%

World-class: >40% Good: 20-40% Typical: 5-15% CURRENT: 0.7%

════════════════════════════════════════════════════════════

Bottlenecks

RankStepWaitRoot CauseAgent Fix
1Follow-up5-14 daysNo system, relies on memorySequencer auto-schedules, CRM tasks with dates
2Find prospect2-3 daysManual search, gets deprioritisedProspect Research Service: company → profile in 60s
3Send + log1-2 daysProcrastination, no urgency signalCompose + schedule in one step, auto-log to CRM
4Assess fit2 daysGut feeling, no modelLead Scoring: ranked list with reasoning
5Write message1 hourBlank page problem, no personalization dataOutreach Composer: draft from research + SPCL gate

Seven Wastes

WasteCurrentWith Agent
Waiting12 days of gaps between stepsSequencer eliminates inter-step delays
ReworkSame bad message sent to everyoneSPCL gate catches quality < 3.5 before send
MotionContext switching between LinkedIn, email, CRMOne workflow: Research → Score → Compose → Sequence
OverproductionMessaging cold prospects equallyRouting: Hot/Warm/Cold get different treatments
Over-processingManual logging of every touchpointAuto-log on send, webhook on open/reply

Target State

  Agent-assisted flow:
Cycle Time: ~15 min (research 5min + review 5min + approve 5min)
Wait Time: ~1 day (human review queue)
Lead Time: ~1.5 days
Flow Efficiency: 15min / 1.5 days ≈ 7% → 10x improvement

At scale (agent handles full queue):
Flow Efficiency target: 20-40%

Gate

Before moving to Dependency Map:

  • All 5 steps documented with cycle time — YES
  • Wait time measured between each step — YES (0 to 14 days)
  • Flow efficiency calculated — YES (0.7% current, 7% target)
  • Bottlenecks identified and ranked — YES (5 ranked)
  • Waste types categorized — YES (5 of 7 present)
  • Improvement opportunities prioritized — YES (agent fixes mapped to each bottleneck)

Context