Product dashboard: DAX, visualization, validation & findings

Live Power BI dashboard

Published report walmart_visualization — Executive, Store, Product, and Demand pages on Gold aggregates. Open in Power BI ↗

Analysis pages: 08 Executive · 09 Store · 10 Product · 11 Demand · Overview

Part A — DAX measures used on Product (Page 3)

Full catalogue: 08 Executive · Part A. Product page measures:

Product Units =
SUM ( 'agg_product_performance'[units_sold] )

Product Sales Value =
SUM ( 'agg_product_performance'[sales_value] )

Product Demand Volatility =
AVERAGE ( 'agg_product_performance'[demand_volatility] )

Product Avg Daily Units =
AVERAGE ( 'agg_product_performance'[avg_daily_units] )

Product Units Rank =
RANKX (
    ALLSELECTED ( 'agg_product_performance'[item_id] ),
    [Product Units],
    ,
    DESC,
    DENSE
)

Part B — Dashboard 3 screenshot

1. Scope & method

2. Visualization & interaction

2.1 Layout

┌──────────────────────────────────────────────────────────────┐
│ [ Product Units ] [ Product Sales $ ] [ Demand Volatility ]  │
│ [ cat_id slicer ]                                            │
│ [ dept_id slicer ]  [ Top 15 bar ]      [ Volatility scatter ]│
│                 [ Sales by cat ]        [ Price analysis tbl ]│
└──────────────────────────────────────────────────────────────┘

2.2 Visual-to-field map

2.3 Interaction design

2.4 Suggested click-path

1. Read KPIs (network volume, sales, average volatility).
2. Scan Top 15 — note all are FOODS_3.
3. On scatter, find large FOODS bubbles at high avg_daily_units (right side).
4. Filter cat_id = HOUSEHOLD → see lower volume, different price profile in table.
5. Clear → select FOODS_3 in dept slicer → confirm Top 15 / scatter focus.
6. Use price table for dept-level pricing bands (esp. HOUSEHOLD_2 max).

3. Key pattern findings

3.1 Top movers are exclusively FOODS_3

All Top 15 SKUs by units are in department FOODS_3. Leader FOODS_3_090: 1,017,916 units (~1.5% of all units), avg daily units 52.4, demand volatility 60.9 (also #1 volatility).

High-velocity FOODS items also tend to be high-volatility — volume and instability travel together for the head of the assortment (corr(avg_daily_units, demand_volatility) ≈ 0.94 across SKUs).

3.2 Category economics

FOODS is both the largest and the most volatile category on average. HOBBIES is smallest and calmest — fewer SKUs, lower unit velocity.

3.3 Price structure by department

FOODS_3 drives the most sales at a low average price — classic high-velocity grocery. HOUSEHOLD_2 has the extreme max price ($107.32) but far less total sales than FOODS_3.

3.4 Scatter interpretation

Most SKUs cluster near the origin (low avg daily units, low volatility). A small FOODS tail extends to the upper-right: high run-rate and high volatility. Those SKUs need tighter forecast / safety-stock attention than the long tail of quiet items.

4. Dashboard quality notes

• ✓ Scatter uses item_id on Values with Average of avg_daily_units / demand_volatility — correct grain (not 3 summed category bubbles).
• ✓ Top 15 filter matches units_rank / unit ranking in Gold.
• ! Format Product Sales Value card like Executive ($191.58M, not $191.57754604M).
• ! Slicers: set Multi-select with CTRL = Off on cat_id / dept_id for easier multi-toggle (same fix as Store page).
• ! Optional: add tooltip fields on Top 15 / scatter (dept_id, avg_sell_price, stores_selling).

5. Recommendations

1. Prioritise demand planning on FOODS_3 head SKUs (especially FOODS_3_090, _586).
2. Treat high avg-daily + high volatility items as forecast-critical; long-tail SKUs can use simpler baselines.
3. Use dept price table for assortment pricing reviews — watch HOUSEHOLD_2 outliers.
4. Format sales card; enable easy multi-select on category/dept slicers.
5. Sync category slicer with Executive/Store if stakeholders drill from FOODS columns.

6. Reproduce

python - <<'PY'
import pandas as pd
from pathlib import Path
P = Path('data/gold_export')
pp = pd.read_parquet(P/'agg_product_performance.parquet')
print(pp['units_sold'].sum(), pp['sales_value'].sum(), pp['demand_volatility'].mean())
print(pp.nsmallest(15, 'units_rank')[['item_id','units_sold','demand_volatility']])
print(pd.read_parquet(P/'agg_price_analysis.parquet').sort_values('sales_value', ascending=False))
PY

Screenshot: assets/product-dashboard-page3.png. Related: 08 Executive · 09 Store.