group-benchmarking

Group Benchmarking — Rooftop-vs-Rooftop Performance Comparison

User Profile (Required)

This skill requires a dealer group profile with at least 2 locations.

1. Read the marketcheck-profile.md project memory file.
2. Check user_type:
   • If dealer_group: use dealer_group.locations[] — requires minimum 2 locations
   • If dealer or other: "This skill is for multi-location dealer groups. Run /onboarding as a Dealer Group."
3. Extract:
   • locations[] with dealer_id, name, zip, state, dealer_type, franchise_brands for each
   • preferences for group defaults
   • country ← location.country
4. Confirm: "Benchmarking [N] locations for [group_name]"

User Context

The primary user is a dealer group executive (CEO, VP Operations, Regional Director) who needs to identify which stores are outperforming and which need intervention. The goal is to surface best practices from top performers and flag underperformers with specific metrics for improvement.

Workflow: Rooftop Benchmarking

Step 1 — Collect KPIs per location

For each location, use the Agent tool to spawn the marketcheck-cowork-plugin:lot-scanner agent in facets mode:

Fetch inventory stats for dealer_id=[dealer_id], country=[country]. Mode: full (with DOM stats) Return: total_units, avg_dom, median_dom, units_under_30_dom, units_30_60_dom, units_over_60_dom, avg_price

From the scanner results, calculate per location:

• Total units on lot
• Avg DOM across all units
• Turn rate = 30 / avg_dom (higher = faster turns)
• Aged % = units_over_60 / total × 100
• Fresh % = units_under_30 / total × 100 (recently acquired)
• Avg listing price

Step 2 — Pricing efficiency (US only)

For each US location, use the Agent tool to spawn the marketcheck-cowork-plugin:lot-pricer agent on a SAMPLE of units (e.g., the 10 oldest and 5 newest per location):

Price these vehicles: [sample VINs with miles and listed_price] zip: [location zip], dealer_type: [location dealer_type]

From results, calculate per location:

• Avg price-to-market % (are they priced competitively?)
• Overpriced % = units with gap > +5% / sampled units × 100
• Underpriced % = units with gap < -5% / sampled units × 100

Step 3 — Market context per location (US only)

For each location, call mcp__marketcheck__get_sold_summary with:

• state: location's state
• inventory_type: Used
• ranking_measure: average_days_on_market
• date_from / date_to: prior month

This gives the LOCAL market average DOM — which provides context for whether a location's DOM is good or bad RELATIVE to its market.

Calculate:

• DOM vs Market = location avg DOM - local market avg DOM (negative = better than market)
• Market-adjusted score = accounts for the fact that some markets are inherently slower

Step 4 — Rank locations on each KPI

Create rankings (1 = best):

1. Turn Rate Rank (highest turn rate = rank 1)
2. Aged % Rank (lowest aged % = rank 1)
3. Pricing Efficiency Rank (closest to market = rank 1)
4. DOM vs Market Rank (most below market = rank 1)
5. Composite Rank = average of all individual ranks

Step 5 — Identify patterns

• Best practices: What is the #1 ranked location doing differently? (e.g., lower avg DOM, tighter pricing, better mix)
• Underperformer root cause: For the lowest-ranked location, identify the primary driver (pricing? aging? mix?)
• Outliers: Any location with a metric > 2 standard deviations from the group mean

Output

GROUP BENCHMARKING — [Group Name]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
[N] locations compared | Period: [Month Year]

PERFORMANCE RANKINGS (1 = best)

Location           | Units | Avg DOM | Turn Rate | Aged % | Price Eff | DOM vs Mkt | Composite
-------------------|-------|---------|-----------|--------|-----------|------------|----------
★ [Location 1]     | XXX   | XX      | X.XX      | X%     | +X.X%     | -X days    | 1
  [Location 2]     | XXX   | XX      | X.XX      | XX%    | +X.X%     | +X days    | 2
  [Location 3]     | XXX   | XX      | X.XX      | XX%    | +X.X%     | +X days    | 3
  ...
⚠ [Location N]     | XXX   | XX      | X.XX      | XX%    | +X.X%     | +X days    | N

★ = Top performer | ⚠ = Needs attention

GROUP AVERAGES
  Avg DOM: XX days | Turn Rate: X.XX | Aged %: XX% | Price-to-Market: +X.X%

KPI DEEP DIVE

Turn Rate:
  Best:  [Location] at X.XX (XX units sold per 30 days equivalent)
  Worst: [Location] at X.XX
  Gap:   X.XX (X.Xx difference — [worst] is turning XXx slower)

Aging:
  Best:  [Location] at X% aged (only X units over 60 days)
  Worst: [Location] at XX% aged (XX units — $X,XXX/day in floor plan)
  Gap:   XX percentage points

Pricing Efficiency:
  Best:  [Location] at +X.X% vs market (tight, competitive pricing)
  Worst: [Location] at +XX.X% (significantly overpriced, contributing to aging)

BEST PRACTICES (from top performer)
- [e.g., "[Location 1] prices within 3% of market on 90% of units — aggressive day-1 pricing prevents aging"]
- [e.g., "[Location 1] has only 4% aged inventory — suggests strong reconditioning-to-frontline speed"]

IMPROVEMENT OPPORTUNITIES (for bottom performers)
- [e.g., "[Location N] has 25% aged inventory costing $X,XXX/day. Immediate action: reduce prices on XX aged units by avg $X,XXX to reach market level"]
- [e.g., "[Location N-1] is priced XX% above market on average. Aligning to market could reduce avg DOM by XX days"]

RECOMMENDED ACTIONS
1. [Most impactful action with specific location, metric, and dollar impact]
2. [Second action]
3. [Third action]

Important Notes

• This skill requires dealer_id for all locations. Locations without a dealer_id are excluded with a note.
• US vs UK: Full benchmarking (including pricing efficiency and market context) for US. UK locations get inventory metrics only (total, DOM, aging) without market pricing comparison.
• For groups with 10+ locations, consider grouping by state or region before ranking.
• The composite rank is a simple average — it weights all KPIs equally. If the user has a specific priority (e.g., "I care most about turn rate"), adjust the weighting.
• Sampling for pricing efficiency (Step 2) uses 15 units per location. For a more thorough analysis, suggest running /weekly-review on the underperforming location.

Self-Check (before presenting to user)

• All locations in the group represented
• Metrics normalized for fair comparison (per-unit basis where appropriate)
• Best and worst performers clearly identified
• Recommendations are location-specific, not generic
• Data period cited