Dataset: 7,200 records × 14 columns | Completeness: 100% | Duplicates: 0
HEALTHY Completeness: 100% complete across all 14 columns and 7,200 records. Zero missing values detected.
HEALTHY Duplicates: Zero duplicate records found. Each record represents a unique SKU-week-channel-region observation.
HEALTHY Data Types: All columns have appropriate data types. Numeric columns (unit_price, unit_cost, discount_rate, forecast_demand_units, opening_inventory_units, units_sold, stockout_flag, lost_revenue_estimate, gross_margin_dollars) correctly typed as float64 or int64. Categorical dimensions (sku, category, channel, region, week) correctly typed as object.
MONITOR SKU Cardinality: 7,200 unique SKUs in 7,200 records indicates each SKU appears exactly once (unique combination of SKU × week × channel × region). High entropy (12.81) confirms extreme diversity. This structure suggests the dataset represents weekly snapshots rather than repeated time series per SKU, limiting longitudinal SKU-level analysis.
MONITOR Outliers - Right-Skewed Distributions: Multiple columns exhibit high right skewness: forecast_demand_units (skewness 4.48, 6.3% outliers), opening_inventory_units (skewness 6.04, 7.5% outliers), units_sold (skewness 4.80, 6.7% outliers), lost_revenue_estimate (skewness 7.65, 14.3% outliers), and gross_margin_dollars (skewness 6.05, 7.5% outliers). These represent legitimate long-tail business patterns (high-volume SKUs, large orders) rather than data errors, but concentration in the tail warrants segmentation analysis.
MONITOR Zero-Inflation: discount_rate is zero for 76.5% of records, and lost_revenue_estimate is zero for 62.8% of records (non-stockout events). stockout_flag is binary with 37.2% positive class prevalence. These are expected structural characteristics, not data quality issues, but create highly non-normal distributions.
HEALTHY Price-Cost Relationship: unit_price and unit_cost show very strong positive correlation (r=0.94), indicating consistent markup practices across the product range without anomalous pricing.
Distribution: Mean $101.12, median $100.61, range $18.05–$215.90. Skewness 0.07 indicates an approximately symmetric distribution centered near $100. Zero outliers detected. The distribution shows even spread across the price range with no concentration in premium or budget segments.
Distribution: Mean $53.47, median $51.69, range $7.25–$140.01. Skewness 0.31 indicates near-symmetry with a slight right tail. Only 1 outlier (0.01%) detected. The cost structure mirrors the price distribution, supporting the strong price-cost correlation and suggesting standardized margin targets.
Distribution: Mean 4.1%, median 0.0%, range 0–30%. Skewness 1.95 indicates high right skew. 76.5% of records have zero discount. The only non-zero values present are 5%, 10%, 15%, 20%, 25%, and 30% — six discrete tiers with no intermediate values. This reflects a rigid promotional structure rather than dynamic pricing. The 23.5% outlier rate reflects the IQR method flagging all non-zero discounts, not true anomalies.
Distribution: Mean 32.2 units, median 23.0 units, range 2–751 units. Skewness 4.48 indicates extreme right skew. 6.3% outliers represent high-velocity SKUs. The median is 28% below the mean, indicating concentration in lower-volume items with a long tail of high-demand SKUs. This distribution pattern is typical for retail inventory but creates challenges for allocation if forecasts do not account for tail behavior.
Distribution: Mean 39.1 units, median 25.0 units, range 1–1,355 units. Skewness 6.04 indicates extreme right skew. 7.5% outliers. The median is 36% below the mean, showing even stronger concentration in low-inventory levels than forecast demand. This suggests systematic under-stocking, particularly given the 37.2% stockout rate. The gap between median opening inventory (25) and median forecast demand (23) is minimal, leaving little safety stock buffer.
Distribution: Mean 28.1 units, median 19.0 units, range 1–751 units. Skewness 4.80 indicates extreme right skew. 6.7% outliers. Sales are inventory-capped when stockouts occur, meaning this distribution reflects observed sales, not true demand. The median is 32% below the mean. The mean units_sold (28.1) is 12.7% lower than mean forecast_demand_units (32.2), consistent with the 37.2% stockout rate reducing realized sales.
Distribution: Binary column with mean 0.37 (37.2% stockout rate). 2,676 stockout events out of 7,200 records. This is a high stockout rate for any retail operation and represents substantial lost revenue opportunity. The prevalence is sufficient for predictive modeling without severe class imbalance issues.
Distribution: Mean $412.22, median $0.00, range $0–$29,880. Skewness 7.65 indicates extreme right skew. 14.3% outliers. The zero median reflects that 62.8% of records have zero lost revenue (no stockout). Among stockout events, mean lost revenue is $1,109.12, indicating significant per-event impact. The extreme skewness shows most stockout events have modest lost revenue, but a tail of high-value missed sales drives total impact.
Distribution: Mean $1,172.37, median $706.02, range $5.73–$41,064.68. Skewness 6.05 indicates extreme right skew. 7.5% outliers. This represents realized margin on sold units only. The distribution closely mirrors units_sold, as expected. The median is 40% below the mean, showing concentration in lower-margin transactions with a long tail of high-value sales.
Distribution: Apparel 24.8%, Home Goods 18.2%, Outdoor 16.3%, Electronics 15.0%, Beauty 14.1%, Pet Supplies 11.6%. Entropy of 2.54 indicates moderate diversity — no single category dominates, but Apparel has 2.1× the share of Pet Supplies. The distribution shows reasonable balance for portfolio analysis, with sufficient sample sizes in each category (minimum 835 records) to support reliable segment rate calculations.
Distribution: Shopify 51.7%, Amazon 24.3%, Retailer Portal 14.1%, Wholesale 9.9%. Entropy of 1.72 indicates moderate concentration. Shopify dominates with over half of all records, reflecting primary direct-to-consumer focus. Wholesale represents the smallest channel by volume but shows the highest stockout rate (39.3%), indicating it may be under-resourced relative to its demand profile.
Distribution: Midwest 20.3%, West 20.1%, Southeast 20.0%, Northeast 20.0%, Southwest 19.6%. Entropy of 2.32 indicates high diversity — nearly perfect geographic balance with maximum 0.7pp spread between regions. This even distribution is atypical for retail (does not track US population distribution) and suggests either deliberate allocation balancing or artificial sampling. The balance ensures regional stockout rate differences are not artifacts of volume disparities.
Distribution: Perfectly balanced across 26 weeks (2026-W01 through 2026-W26), each representing 3.85% of records (277 records per week, with Week 26 at 276). Entropy of 4.70 indicates very high diversity. This represents a 6-month observation period with systematic weekly snapshots. The temporal balance enables clean week-over-week comparison without volume bias.
Distribution: Maximum entropy — each SKU appears exactly once. This indicates the dataset structure is SKU × week × channel × region unique combinations rather than repeated observations per SKU. The high cardinality prevents SKU-level trend analysis but supports segment-level pattern detection across categories and channels.
Opening inventory and units sold move together almost perfectly. This correlation has two interpretations depending on stockout status: (1) for non-stockout records, it reflects successful demand fulfillment where higher inventory enables higher sales, and (2) for stockout records, it reflects inventory-constrained sales where low inventory caps realized sales. The very strong correlation confirms that inventory availability is the primary constraint on sales volume — not demand variability. Operationally, this means improving forecast accuracy alone will not increase sales unless matched with higher inventory allocation.
Censored Sales Warning: This correlation reflects inventory-capped observed sales matching forecast demand, not forecast accuracy. When stockouts occur (37.2% of records), units_sold is capped by opening_inventory_units regardless of true demand. The correlation tells us that inventory allocation closely follows forecasts (as expected), but it does not validate forecast quality. True forecast accuracy cannot be assessed from this dataset because actual unconstrained demand is not observable when stockouts occur. Any claim that "forecasts are accurate" based on this correlation would be incorrect.
Price and cost track together almost perfectly, indicating standardized markup practices across the product range. The consistent relationship suggests either cost-plus pricing or parallel sourcing/pricing strategies. This is operationally healthy — it means pricing errors or margin compression are unlikely. The correlation also means price and cost provide redundant information for predictive modeling; only one needs to be included as a feature.
Forecasted demand and opening inventory are strongly aligned, indicating that the replenishment system uses forecasts as the primary input for inventory allocation. The correlation being slightly weaker than forecast-to-sales (0.86 vs. 0.95) suggests either (1) inventory constraints (cannot always allocate forecast quantity) or (2) deliberate under-allocation to manage working capital risk. The 37.2% stockout rate implies the latter — opening inventory is systematically set below forecast demand, creating planned shortfalls.
Sales volume drives margin dollars, as expected. The correlation is strong but not perfect because margin per unit varies by product mix (price, cost, discount rate). The relationship confirms that increasing sales volume (by reducing stockouts) would directly increase total margin capture proportionally.
Stockout events correlate moderately with lost revenue magnitude. The correlation is not stronger because lost revenue per stockout event varies widely based on unmet demand quantity and unit price. A stockout of a high-priced, high-demand SKU generates much higher lost revenue than a low-priced, low-demand SKU. This variability is why segment analysis by category and channel is critical — not all stockouts have equal business impact.
Highest stockout rate: Northeast with 39.6% (2.4pp above overall 37.2%). Lowest stockout rate: Southeast with 35.4% (1.8pp below overall).
Largest total lost revenue: Northeast leads with $674K in lost revenue across 569 stockout events, representing 22.7% of total company lost revenue despite being 20.0% of records. Midwest follows with $588K (540 events), West with $582K (535 events), Southeast with $574K (511 events), and Southwest with $550K (521 events).
Northeast's combination of highest rate and highest total impact makes it the priority region for inventory reallocation. The 4.2pp gap between highest (Northeast 39.6%) and lowest (Southeast 35.4%) region stockout rates indicates meaningful regional allocation differences that are addressable through redistribution.
Highest stockout rate: Wholesale with 39.3% (2.2pp above overall 37.2%). Lowest stockout rate: Shopify with 35.7% (1.4pp below overall).
Largest total lost revenue: Shopify accounts for $1.55M in lost revenue (52.1% of total) across 1,329 stockout events, reflecting its 51.7% share of total records and dominant volume. Amazon follows with $677K (673 events), Retailer Portal with $459K (394 events), and Wholesale with $285K (280 events).
Wholesale shows the highest stockout rate at 39.3% despite being only 9.9% of volume, indicating systematic under-allocation relative to its demand profile. Shopify has the lowest rate at 35.7% and captures the largest absolute lost revenue due to volume dominance. The 3.6pp gap between Wholesale (39.3%) and Shopify (35.7%) stockout rates represents an opportunity to rebalance allocation, though Shopify's larger total lost revenue means even modest rate improvements there yield high absolute dollar recovery.
Highest stockout rate: Outdoor with 38.6% (1.4pp above overall 37.2%). Lowest stockout rate: Pet Supplies with 36.1% (1.1pp below overall).
Largest total lost revenue: Apparel leads with $742K in lost revenue across 662 stockout events (25.0% of total company lost revenue), despite a near-average 37.1% stockout rate. The high total impact is driven by Apparel's 24.8% volume share. Home Goods follows with $530K (490 events), Electronics with $539K (391 events), Outdoor with $460K (454 events), Beauty with $380K (378 events), and Pet Supplies with $317K (301 events).
Outdoor shows the highest stockout rate at 38.6% (1.4pp above overall) but ranks fourth in total lost revenue. Apparel combines near-average rate with highest total impact due to volume dominance, making it the highest-priority category by business impact. The 2.5pp gap between highest (Outdoor 38.6%) and lowest (Pet Supplies 36.1%) category stockout rates is modest, indicating category-level allocation is more balanced than channel or region.
1. Apparel × Shopify: 35.5% stockout rate (2.6pp below overall), $376K lost revenue across 322 events. Largest absolute lost revenue of any combination due to high volume (931 records). Rate is below average, indicating Shopify's Apparel inventory is better allocated than other segments.
2. Electronics × Shopify: 35.2% stockout rate (2.0pp below overall), $317K lost revenue across 189 events. Second-highest lost revenue. Below-average rate shows relatively strong allocation.
3. Home Goods × Shopify: 38.1% stockout rate (0.9pp above overall), $297K lost revenue across 269 events. Third-highest lost revenue. Modestly above-average rate suggests minor allocation gap.
4. Beauty × Shopify: 36.0% stockout rate (1.2pp below overall), $210K lost revenue across 197 events. Below-average rate.
5. Outdoor × Shopify: 35.2% stockout rate (2.0pp below overall), $189K lost revenue across 204 events. Below-average rate.
6. Apparel × Amazon: 40.9% stockout rate (3.8pp above overall), $176K lost revenue across 176 events. Above-average rate indicates Apparel is under-allocated on Amazon relative to Shopify.
7. Pet Supplies × Shopify: 35.4% stockout rate (1.8pp below overall), $158K lost revenue across 148 events. Below-average rate.
8. Apparel × Retailer Portal: 39.5% stockout rate (2.3pp above overall), $137K lost revenue across 105 events. Above-average rate.
9. Outdoor × Amazon: 42.9% stockout rate (5.7pp above overall), $124K lost revenue across 129 events. Second-highest stockout rate among top combinations, indicating meaningful allocation gap for Outdoor on Amazon.
10. Outdoor × Retailer Portal: 46.8% stockout rate (9.6pp above overall), $116K lost revenue across 80 events. Highest stockout rate among all combinations, indicating severe under-allocation for Outdoor via Retailer Portal despite modest absolute lost revenue.
Key Cross-Tab Insights: Shopify consistently shows below-average stockout rates across most categories, indicating it receives priority allocation. Amazon and Retailer Portal show elevated rates for Outdoor (42.9% and 46.8% respectively) and Apparel (40.9% on Amazon), indicating these category-channel combinations are systematically under-allocated. The Outdoor × Retailer Portal combination's 46.8% rate (9.6pp above overall) represents the single highest-rate segment in the dataset and warrants immediate inventory reallocation despite ranking 10th by total lost revenue.
Effect size analysis identifies which features have the strongest relationship with stockout events, distinguishing statistically significant patterns from practically meaningful predictors.
Opening inventory shows a medium negative effect on stockouts. Stockout events have 20.8 units opening inventory on average, compared to 50.0 units for non-stockout events — a 58% lower starting position. This is the strongest operational predictor of stockouts in the dataset. The negative direction confirms that low initial inventory is the primary driver of stockouts, not demand surges. Allocating higher opening inventory to high-risk segments would directly reduce stockout frequency.
Lost revenue shows a large positive effect, but this is definitional rather than predictive — stockout events by definition have non-zero lost revenue ($1,109 mean), while non-stockout events have zero. This statistic confirms data integrity but does not provide actionable insight beyond the obvious.
Units sold shows a small negative effect on stockouts. Stockout events average 20.8 units sold vs. 32.5 units for non-stockout events — 36% lower. This reflects inventory-capped sales during stockouts but is less informative than opening inventory because it is an outcome rather than a leading indicator.
Realized gross margin shows a small negative effect. Stockout events capture $869 in margin on average vs. $1,352 for non-stockout events — 36% lower. This reflects the opportunity cost of stockouts but is derivative of units_sold and does not add independent predictive value.
Price, cost, discount rate, and forecast demand all show negligible effect sizes (Cohen's d between -0.02 and 0.03). This indicates stockouts are not meaningfully correlated with product price tier, discount status, or forecast demand level — stockouts occur across the full range of these variables. The lack of forecast_demand_units effect (d = -0.02) is particularly notable: it suggests stockouts are not concentrated in high-forecast SKUs, reinforcing that the issue is allocation level (opening_inventory_units) rather than demand pattern recognition.
For a stockout risk classifier, opening_inventory_units is the single strongest feature (medium effect). Segment categorical features (category, channel, region, week) should be included to capture allocation policy differences. Price, cost, and discount features can be excluded as they show negligible effect sizes. Forecast demand is weakly predictive alone but may have interaction effects with opening inventory (forecasted vs. allocated ratio) that would emerge in a tree-based model.
1. Prioritize Northeast Region Inventory Reallocation
Evidence: Northeast has a 39.6% stockout rate (2.4pp above company average) and accounts for $674K in lost revenue (22.7% of total) across 569 stockout events, the highest total lost revenue of any region despite representing 20.0% of records.
Impact: High. Reducing Northeast stockout rate to the Southeast level (35.4%) would eliminate approximately 60 stockout events per comparable period.
Difficulty: Medium. Requires cross-regional inventory rebalancing and potentially revised allocation rules, but no infrastructure changes.
First Step: Audit the allocation logic for Northeast vs. Southeast to identify the systematic difference driving the 4.2pp rate gap, then pilot a 10% opening inventory increase for Northeast's top 20 SKUs by forecast demand.
2. Increase Wholesale Channel Opening Inventory Buffer
Evidence: Wholesale has the highest stockout rate of any channel at 39.3% (2.2pp above overall), despite representing only 9.9% of volume. Opening inventory for Wholesale stockout events averages 20.8 units, identical to the overall stockout mean, indicating Wholesale receives no allocation premium despite higher risk.
Impact: Medium. Wholesale's $285K lost revenue is material, and the 3.6pp gap vs. Shopify (35.7%) is addressable.
Difficulty: Easy. Wholesale is a small channel by volume (712 records), making targeted inventory increases operationally simple.
First Step: Set opening inventory for Wholesale SKUs to 110% of forecast demand (vs. current ~100%), monitor stockout rate change over 4 weeks, and adjust multiplier if needed.
3. Address Outdoor × Retailer Portal Combination (Highest Stockout Rate)
Evidence: Outdoor × Retailer Portal shows a 46.8% stockout rate (9.6pp above overall) with $116K lost revenue across 80 events. This is the single highest stockout rate of any category-channel combination and indicates severe under-allocation.
Impact: Medium-High. The absolute lost revenue ranks 10th, but the extreme rate gap (9.6pp) signals a fixable allocation error.
Difficulty: Easy. This is a small, well-defined segment (171 total records) amenable to targeted intervention.
First Step: Increase opening inventory for Outdoor SKUs routed through Retailer Portal by 20%, and implement a minimum 3-week forward cover policy for this combination to reduce volatility risk.
4. Focus Apparel Allocation on Amazon and Retailer Portal
Evidence: Apparel drives the highest total lost revenue ($742K, 25.0% of company total), but stockout rates vary by channel: Shopify Apparel has a 35.5% rate (below average), while Amazon Apparel has 40.9% (3.8pp above overall) and Retailer Portal Apparel has 39.5% (2.3pp above overall). This indicates Shopify receives priority Apparel allocation while other channels are under-served.
Impact: High. Apparel's volume dominance means even modest rate reductions yield large absolute recovery.
Difficulty: Medium. Requires channel-specific allocation logic changes but no product changes.
First Step: Rebalance Apparel opening inventory to equalize stockout rates across channels — reduce Shopify Apparel allocation by 5% and redirect to Amazon and Retailer Portal to close the 5.4pp gap between Shopify and Amazon Apparel rates.
5. Use Opening Inventory as Primary Allocation Lever
Evidence: Opening inventory shows a medium effect size (Cohen's d = -0.66) on stockout risk. Stockout events have 58% lower opening inventory (20.8 units vs. 50.0 units). The very strong correlation between opening inventory and units sold (r = 0.95) confirms inventory availability is the binding constraint on sales, not demand variability.
Impact: High. Opening inventory is the single most predictive feature of stockout risk and the most direct operational lever.
Difficulty: Medium. Requires revised allocation algorithms and potentially higher working capital, but no process redesign.
First Step: Establish segment-specific opening inventory targets as a multiple of forecast demand (e.g., Wholesale 1.1x, Retailer Portal 1.15x, Shopify 1.0x) and implement as allocation policy for the next replenishment cycle.
6. Audit Week 13, Week 20, and Week 7 for Seasonal Under-Allocation
Evidence: Week 13 has the highest stockout rate at 42.6% (5.4pp above overall) with $116K lost revenue. Week 20 shows 42.2% (5.1pp above overall) with $113K lost revenue. Week 7 shows 41.2% (4.0pp above overall) with $132K lost revenue. In contrast, Week 9 has the lowest rate at 33.2% (4.0pp below overall). This 9.4pp range between highest and lowest weeks indicates systematic under-allocation during specific periods.
Impact: Medium. Temporal patterns are addressable if the cause is identified (e.g., lead time lag, seasonal demand spike, post-promotion depletion).
Difficulty: Medium. Requires root cause analysis of high-stockout weeks and potentially revised replenishment timing.
First Step: Compare forecast accuracy and lead time performance for Weeks 13, 20, and 7 vs. low-stockout weeks to determine whether the issue is demand prediction, replenishment timing, or allocation level.
7. Implement Segment-Specific Safety Stock Policies
Evidence: Median opening inventory is 25.0 units vs. median forecast demand of 23.0 units — only 2 units of buffer, or 8.7% safety stock. Given 37.2% stockout rate, this buffer is insufficient. The correlation between forecast and opening inventory (r = 0.86) shows allocation follows forecasts closely but without adequate safety margin for high-risk segments.
Impact: High. Insufficient safety stock is a root cause of the high baseline stockout rate.
Difficulty: Medium. Requires segment-specific safety stock rules and higher working capital investment.
First Step: Set minimum safety stock levels by segment risk tier: 20% buffer for Wholesale and Retailer Portal, 15% for Amazon, 10% for Shopify, applied as a floor on opening inventory allocation.
8. Standardize Discount Rate Application to Manage Overstock Risk
Evidence: 76.5% of records have zero discount, and the six discrete discount tiers (5%, 10%, 15%, 20%, 25%, 30%) are applied to only 23.5% of inventory. Cohen's d = 0.01 shows discount rate has negligible effect on stockout risk, indicating discounts are not used to clear overstock proactively or shift demand away from stockout-prone SKUs.
Impact: Low-Medium. Dynamic discounting is a demand-side lever that complements supply-side allocation improvements.
Difficulty: Easy. Promotional infrastructure exists; this is a policy change to expand usage.
First Step: Pilot dynamic discounting for overstock SKUs in categories where stockout risk is low (e.g., Pet Supplies with 36.1% rate) to shift demand and fund higher opening inventory for stockout-prone categories (e.g., Outdoor, Wholesale channel).
9. Develop a Gradient Boosting Stockout Risk Model (Advanced)
Evidence: The dataset has complete, clean data with 37.2% positive class prevalence (no severe imbalance), strong predictive features (opening_inventory_units Cohen's d = -0.66), and categorical segments with measurable rate differences. Gradient Boosting is well-suited to capture non-linear interactions (e.g., category × channel, opening_inventory × forecast_demand ratio).
Impact: Medium. A trained model would enable SKU-level risk scoring for proactive allocation adjustments.
Difficulty: Medium. Requires model training, validation, and integration into allocation workflow.
First Step: Train a Gradient Boosting classifier using opening_inventory_units, forecast_demand_units, category, channel, region, and week as features. Validate on a holdout set and deploy risk scores to flag high-risk SKU-week combinations for increased opening inventory allocation.
10. Conduct SKU-Level Analysis for High Lost Revenue Events
Evidence: Lost revenue distribution is highly right-skewed (skewness 7.65) with 14.3% outliers, including a maximum event of $29,880. These tail events represent disproportionate business impact but are not addressable through segment-level recommendations alone. Each SKU appears exactly once in the dataset, so time-series SKU patterns are not observable here.
Impact: Medium. Preventing even a few high-impact stockouts per period would materially reduce total lost revenue.
Difficulty: Medium. Requires longitudinal SKU data and SKU-specific allocation rules.
First Step: Identify the top 50 SKUs by lost_revenue_estimate in this dataset and request historical replenishment data for those SKUs to determine whether high lost revenue events are predictable from prior cycles or represent one-time shocks.
HIGH SEVERITY Inventory-Capped Sales Limiting Demand Visibility
The very strong correlation between forecast_demand_units and units_sold (r = 0.95) cannot be interpreted as forecast accuracy because units_sold is capped by opening_inventory_units when stockouts occur (37.2% of records). True unconstrained demand is not observable in this dataset. This means the business is making allocation decisions based on censored sales history, which systematically underestimates actual demand and perpetuates under-allocation.
Mitigation: Implement demand capture during stockout events — e.g., track waitlist signups, backorders, or cart abandonment data to estimate latent demand when inventory is unavailable. Use this uncensored demand signal to retrain forecast models and avoid the censored sales feedback loop.
HIGH SEVERITY Systematic Under-Allocation Relative to Forecast
Median opening inventory (25.0 units) is only 8.7% above median forecast demand (23.0 units), providing minimal safety stock buffer. The 37.2% stockout rate confirms this buffer is insufficient. Cohen's d = -0.66 for opening_inventory_units shows that low initial inventory is the primary driver of stockouts. This indicates a deliberate or systematic policy of allocating at or below forecast demand, which guarantees high stockout rates in any environment with demand or supply volatility.
Mitigation: Immediately implement segment-specific safety stock policies (Recommendation 7) to raise opening inventory to 110–115% of forecast demand for high-risk segments. Conduct a working capital impact analysis to quantify the cost of carrying higher inventory vs. the $2.97M in lost revenue observed over this 26-week period.
MEDIUM SEVERITY Wholesale Channel Allocation Bias
Wholesale has the highest stockout rate (39.3%, 2.2pp above overall) despite representing only 9.9% of volume. This suggests Wholesale is either deprioritized in allocation decisions or subject to longer lead times / lower service level targets than other channels. If Wholesale represents B2B customers or large retail partners, chronic stockouts risk damaging strategic relationships and losing shelf space.
Mitigation: Audit allocation policy to determine if Wholesale is subject to different service level agreements or lower inventory priority. If so, revise policy to equalize stockout risk across channels unless business strategy explicitly accepts higher Wholesale stockout rates. Implement Recommendation 2 to increase Wholesale opening inventory buffer immediately.
MEDIUM SEVERITY Regional Allocation Imbalance
Northeast region has a 39.6% stockout rate (2.4pp above overall) and accounts for $674K lost revenue (22.7% of total), while Southeast has a 35.4% rate (1.8pp below overall). The 4.2pp gap between highest and lowest regions is not explained by population share (regions are balanced at ~20% each) or known geographic demand patterns. This indicates either (1) Northeast receives systematically lower opening inventory allocation, or (2) Northeast has longer replenishment lead times or higher demand volatility that is not accounted for in allocation logic.
Mitigation: Investigate root cause of Northeast's elevated stockout rate vs. Southeast. If it is allocation level, implement Recommendation 1 to rebalance opening inventory. If it is lead time or volatility, adjust safety stock policy for Northeast specifically to compensate for higher risk factors.
| Column | Mean | Median | Std Dev | Min | Max | Skewness | Outliers |
|---|---|---|---|---|---|---|---|
| unit_price | 101.12 | 100.61 | 48.88 | 18.05 | 215.90 | 0.07 | 0 (0.0%) |
| unit_cost | 53.47 | 51.69 | 27.62 | 7.25 | 140.01 | 0.31 | 1 (0.01%) |
| discount_rate | 0.041 | 0.0 | 0.085 | 0.0 | 0.3 | 1.95 | 1,692 (23.5%) |
| forecast_demand_units | 32.17 | 23.0 | 32.11 | 2.0 | 751.0 | 4.48 | 450 (6.25%) |
| opening_inventory_units | 39.15 | 25.0 | 46.62 | 1.0 | 1355.0 | 6.04 | 541 (7.51%) |
| units_sold | 28.13 | 19.0 | 29.46 | 1.0 | 751.0 | 4.80 | 480 (6.67%) |
| stockout_flag | 0.37 | 0.0 | 0.48 | 0.0 | 1.0 | 0.53 | 0 (0.0%) |
| lost_revenue_estimate | 412.22 | 0.0 | 1140.03 | 0.0 | 29880.0 | 7.65 | 1,032 (14.33%) |
| gross_margin_dollars | 1172.37 | 706.02 | 1527.84 | 5.73 | 41064.68 | 6.05 | 538 (7.47%) |
End of Report | Generated from 7,200 records across 14 columns | 26-week observation period (2026-W01 through 2026-W26)