FN-001 — Ghost Inventory Pattern

Snapshot

• What was found: Wine inventory on paper exceeded physical bottle count by 127 units ($4,318 retail value).
• Why it mattered: Food/bev cost variance consistently 3.2 points above target with no identifiable waste.
• What it cost: $28,400 annually in untracked shrinkage (12.7% of wine program revenue).

---

Signals (Observable Indicators)

1. Cost variance persistent above 2.5 points for 90+ days without seasonal explanation
2. MarginEdge inventory counts reconcile to invoices but not to POS sales velocity
3. Wine-by-the-glass pours per bottle below expected yield (22 pours vs. expected 25-26)
4. Server void rates on wine SKUs 2.1x higher than food voids
5. Weekend shift managers approve voids without inventory adjustment
6. BevSpot par levels unchanged despite velocity shifts over 18 months
7. Physical bottle count delays exceed 45 minutes (suggests avoidance behavior)

---

Evidence Trail

Systems Compared

Toast POS	MarginEdge	BevSpot	Physical Count
1,847 units sold (90d)	1,974 units received (90d)	284 units on-hand	157 units on-hand
$47,238 wine revenue (90d)	$22,116 COGS (90d)	Par: 280 units	Variance: -127 units
46.8% theoretical cost	3.2pt variance	Last updated: 47d ago	Audit date: Nov 12

What Mismatched

• POS vs. Physical: 127-unit gap between theoretical inventory (284) and actual count (157)
• Void Pattern: 34 wine voids in 90 days; 29 occurred Fri-Sun; 26 approved by same 2 managers
• Pour Variance: Actual yield 22.3 pours/bottle vs. expected 25.8 (13.6% deviation)

Artifacts Examined

• Toast POS transaction logs (June 1 - Nov 12, 2025)
• MarginEdge receiving history and inventory snapshots
• BevSpot par level configurations (last modified Feb 2024)
• Physical count worksheet (sommelier-conducted, Nov 12, 2025)
• Server void approval logs (90-day export)

---

Root Cause Hypothesis (Ranked)

1. Insufficient Void Controls + Manager Permission Drift

Confidence: High

Evidence:

• 29 of 34 voids (85.3%) occurred on weekend shifts with manager-on-duty approval
• Same 2 managers approved 76% of all wine voids (compared to 31% of food voids)
• No physical inventory adjustment triggered by void workflow in Toast
• Managers promoted from server roles 8-10 months ago; permissions never tightened post-promotion

2. Uncalibrated Pour Variance (Training Drift)

Confidence: Medium

Evidence:

• 13.6% deviation from expected pour yield indicates over-pouring or spillage
• No refresher training documented since opening (22 months ago)
• 3 of 7 bartenders hired in past 6 months; onboarding checklist incomplete
• Jigger use inconsistent (observed during site visit)

3. Stale Par Levels + Velocity Mismatch

Confidence: Medium

Evidence:

• BevSpot par levels unchanged since Feb 2024 (9 months stale)
• Menu mix shifted significantly in summer (rosé velocity up 40%, reds down 18%)
• Dead stock accumulating (11 SKUs with zero movement in 60+ days, $1,847 tied up)
• Over-ordering on slow SKUs driven by outdated pars

---

Fix Protocol

1. Implement Void → Inventory Adjustment Workflow

• Who: Toast admin + GM
• What: Configure Toast to require inventory adjustment selection on all wine voids; create “Spillage,” “Incorrect Order,” “Comp” categories with mandatory inventory deduction
• Output: Void approvals automatically decrement MarginEdge inventory (tested in staging before rollout)

2. Restrict Weekend Manager Void Permissions

• Who: Owner + Toast admin
• What: Require GM or owner approval for wine voids above $35 retail; weekend managers limited to $35 threshold
• Output: Permission matrix updated in Toast; alert sent to GM for all high-value voids within 30 minutes

3. Recalibrate Pour Standards + Retrain Team

• Who: Sommelier + bar lead
• What: Conduct 90-minute pour calibration session; issue fresh jiggers; post laminated pour cards at each station; spot-check 2 pours/shift for 3 weeks
• Output: Pour variance target <5%; sommelier sign-off on calibration completion per bartender

4. Update BevSpot Pars to Current Velocity

• Who: Sommelier + GM
• What: Pull 90-day sales velocity report from Toast; recalculate pars based on 2-week turnover target; flag dead stock for discount/86
• Output: Updated par levels in BevSpot; dead stock decision made (discount, comp, or donate) within 10 days

5. Institute Weekly Spot Counts (High-Value SKUs)

• Who: Sommelier or bar lead
• What: Count top 15 wine SKUs by revenue every Monday AM; compare to expected inventory based on weekend POS sales; investigate variances >5%
• Output: Spot count log (shared spreadsheet); variance alerts escalated to GM same-day if threshold exceeded

---

Prevention Controls

Permission Controls

• Managers: max $35 wine void approval (above threshold escalates to GM)
• Bartenders: cannot approve own voids
• Servers: no void permissions on wine (must request manager approval)
• Owner/GM: unrestricted, but receive digest of all voids >$35 within 24h

Mapping Rules

• All wine voids must map to inventory adjustment reason code (Spillage, Incorrect Order, Comp, Breakage)
• Reason code auto-decrements MarginEdge/BevSpot inventory (no manual entry)
• “Comp” voids require customer name in notes field (enforced by Toast validation)

Threshold Alerts

• Wine cost variance >2.5 points for 14 consecutive days → alert GM + owner
• Void rate on wine SKUs >8% of category revenue → alert GM + sommelier
• Pour yield deviation >8% on spot-check → mandatory retraining within 72h
• BevSpot par levels unchanged >60 days → alert sommelier to review

Scheduled Reconciliations

• Weekly: Spot count (top 15 SKUs) vs. POS expected inventory
• Monthly: Full physical count vs. MarginEdge on-hand
• Quarterly: BevSpot par level review + dead stock purge

---

Recovery Math

Inputs

Variable	Low	Expected	High	Source
Current shrinkage rate	11.2%	12.7%	14.1%	90d POS vs. physical variance
Post-fix shrinkage rate	2.5%	3.8%	5.0%	Industry benchmark (controlled pour + tight voids)
Wine program revenue	$220K	$234K	$248K	Trailing 12mo actuals annualized
Implementation timeframe	6 weeks	7 weeks	9 weeks	Assumes 2-week training lag + 1-week system config

Calculation

Current annual shrinkage = Revenue × Current shrinkage rate
  = $234,000 × 12.7% = $29,718

Target annual shrinkage = Revenue × Post-fix shrinkage rate
  = $234,000 × 3.8% = $8,892

Annual recovery = Current shrinkage - Target shrinkage
  = $29,718 - $8,892 = $20,826

Range

• Conservative (P10): $17,900/year (assumes low revenue, high residual shrinkage)
• Expected (P50): $20,800/year (assumes expected revenue, moderate tightening)
• Optimistic (P90): $27,400/year (assumes high revenue, aggressive shrinkage reduction to 2.5%)

Assumptions

1. Revenue remains flat (no growth assumed; purely defensive recovery)
2. Void controls reduce discretionary comps by 60% within 4 weeks of rollout
3. Pour calibration reduces over-pour waste by 50% within 3 weeks of training
4. No staff turnover during implementation window (6-8 weeks)
5. BevSpot par updates reduce dead stock tie-up by $1,200 within 60 days (one-time cash benefit, not recurring)

---

What This Is NOT

• NOT a staffing problem (turnover normal; training gap, not talent gap)
• NOT a theft allegation (no evidence of malicious intent; permission drift + training erosion)
• NOT a technology problem (systems work fine; workflow mapping is the issue)
• NOT a vendor overbilling issue (invoices reconcile; problem is post-receipt shrinkage)

---

Context

• Revenue Band: $1.8M-$2.2M annual (full-service, wine-focused)
• Concept Type: Modern American, upscale-casual
• Location Density: Urban downtown district, tourism + local mix
• Stack in Use: Toast POS, MarginEdge (inventory/COGS), BevSpot (ordering/pars), QuickBooks (accounting)
• Concept Age: 22 months post-launch
• Team Size: 24 FTE equivalents (7 FOH, 9 BOH, 3 bar, 5 mgmt/admin)