Brainforge Knowledge

LMNT_Snowflake_Tables_and_Shivani_Questions

7 min read

LMNT: Snowflake Tables & Shivani’s Top Questions (Past 1-2 Weeks)

Date Created: 2026-02-16
Purpose: Quick reference for Robert’s Monday meeting with Shivani

📊 Snowflake Tables Relevant to Discuss

Raw Data Sources

1. External Share (from Amazon/Emerson)

  • Location: External share (not in normal raw database)
  • Source: Amazon shares retail data via Emerson
  • Tables:
    • 12 tables from Emerson for Walmart (POS data)
    • 3 tables for Target (POS data)
  • Note: These live outside the normal raw database structure because they come via external share

2. Raw Database: Shopify Data

  • Location: RAW.POLYTOMIC.SHOPIFY
  • Tables:
    • orders - Contains both e-commerce and wholesale orders
      • Key field: tags field (string) - wholesale orders are tagged with “wholesale”
    • order_line_items
    • customers
  • Note: Wholesale orders come from Shopify but are identified via the tags field

3. Google Sheets Integration

  • Source: Wholesale team CRM lives in Google Sheets
  • Usage: Combined with Shopify data in wholesale_dm_customer model

Modeled Data (Broadmarts)

Wholesale Mart

  • Location: BROADMARTS database
  • Tables:
    • wholesale_mart - Weekly views for wholesale
    • wholesale_dm_customer - Combines Shopify orders + Google Sheets CRM data
      • Compiles all wholesale partners/customers
      • Includes data from both Shopify and their CRM

Retail Mart

  • Location: BROADMARTS database
  • Tables:
    • fact_sales - POS data for retail
      • Contains sales from both Walmart and Target
      • This is the primary POS (point of sale) data
    • retail_fact_walmart_only_sales - Walmart-only data
      • Contains Walmart order management system data (NOT POS)
      • Separate from POS data
      • Important: For total retail revenue, you need to join these two tables together

Products Table

  • Used to join with sales data to create retail models
  • Referenced when joining Walmart and Target data together

Key Technical Notes

  1. Walmart Data Structure:

    • Walmart sends TWO types of data:
      • POS data (in fact_sales with Target)
      • Order management system data (in retail_fact_walmart_only_sales separately)
    • To get total Walmart revenue: Must join fact_sales + retail_fact_walmart_only_sales

  2. Target Data Structure:

    • Only POS data
    • Lives in fact_sales table

  3. Current Modeling Status:

    • ✅ Wholesale mart complete
    • ✅ Retail mart created (but needs joining Walmart tables for total revenue)
    • ⚠️ In Progress: Awaish is working on joining the two Walmart tables together for total retail revenue
    • ❌ E-commerce data NOT yet modeled (only wholesale and retail)

❓ Shivani’s Top Questions (Past 1-2 Weeks)

Retail Data Questions (Most Frequent)

  1. “Have you been able to join any Walmart and Target data?”

    • Context: She wants to know if the two sources are combined
    • Answer: Yes, they’re joined in fact_sales for POS data, but Walmart has additional data in separate table

  2. “What were sales across those two stores yesterday? Would you be able to find it?”

    • Context: She’s testing if basic queries work
    • Answer: Yes, but need to join both Walmart tables for complete picture

  3. “What are you actually seeing in the delta between the Target and Walmart data right now?”

    • Context: She wants to understand differences between sources
    • Follow-up questions she’s asking:
      • Are they defining sales differently?
      • Is point of sales apples to apples?
      • What are the differences in how they talk about “date of the week ending”?

  4. “How is Target performing versus Walmart right now?”

    • Context: Basic business question she can’t answer
    • Status: Should be answerable once tables are properly joined

  5. “What about Target in the Northeast versus Target in California?”

    • Context: Geographic segmentation question
    • Status: Depends on whether geographic data is in the tables

  6. “What can I actually glean from retail data right now?”

    • Context: Frustration that data exists but isn’t actionable
    • Her concern: “Everybody’s like, we have this retail data, but what can I actually glean from it right now?”

Data Access & Technical Questions

  1. “Where’s that orders table?” (Asked Robert directly on call)

    • Context: She asked Robert to pull up orders table, he couldn’t access it
    • Issue: Access/permissions problem in Snowflake
    • Status: Needs to be resolved before Monday meeting

  2. “How are Walmart and Target defining revenue differently?”

    • Context: She’s concerned about reconciliation
    • Her thinking: “Maybe Walmart recognizes revenue at a different time than Target”
    • Her concern: When you look at each platform separately, revenue might be different than what’s shown in the model

Wholesale Revenue Questions

  1. “Why doesn’t our wholesale revenue match finance?”

    • Context: Finance pulled their own reports from accounting system (QuickBooks/NetSuite)
    • Status: Amber is working on reconciliation
    • Discrepancy: Less than 10% difference, but Shivani is concerned
    • Note: Wholesale team doesn’t have issues with the data - only finance

  2. “How do we define net revenue?”

    • Context: She wants clarity on definitions
    • Her questions:
      • Is it revenue minus discounts?
      • Minus refunds?
      • What’s the exact calculation?

Process & Workflow Questions

  1. “Can we just ship me all the orders?” (Recent request)

    • Context: She’s in “sprint mode” and wants raw data quickly
    • Uttam’s response: “I can’t get anything in front of you until it’s perfectly formatted, because that’s what you asked me for. So is that off the table now?”
    • Tension: She wants speed, but previously demanded quality/formatting

  2. “What’s the next version of this going to look like?”

    • Context: She’s asking about retail reporting
    • Her frustration: “I don’t really know what’s been happening on the retail side on your ends”
    • She wants: Clear visibility into progress, not just status updates

🎯 What Shivani Really Wants (Based on Patterns)

Immediate Needs:

  1. Answer basic business questions about retail performance (Walmart vs Target, by region, by category)
  2. Understand data differences between sources (Walmart vs Target definitions, timing, etc.)
  3. Reconcile discrepancies with finance data
  4. Access to data - be able to pull up tables and query them herself

Underlying Frustrations:

  1. “I need something to jam about” - She wants concrete topics, not vague status updates
  2. “If we’re gonna jam, then I need a topic at hand” - She needs specific data questions to explore
  3. “What is gonna make all of this work successful is if it lands for the people at LMNT to be able to answer basic questions” - She’s focused on end-user value, not technical progress

Her Working Style:

  • Wants to validate data live with you (not wait for offline QA)
  • Wants to compare your data to her internal sources (finance, Shopify, etc.)
  • Wants to understand differences rather than just accept the numbers
  • Doesn’t have a structured QA process - that’s why she wants working sessions

💡 Recommended Talking Points for Monday

1. Retail Data Status

  • ✅ Walmart and Target POS data joined in fact_sales
  • ⚠️ Walmart has additional order management data in separate table
  • 🔄 Working on joining both Walmart tables for total revenue
  • 📊 Can answer: Walmart vs Target performance (once joined)
  • ❓ Geographic segmentation: Need to check if available in tables

2. Data Definitions & Differences

  • Walmart sends 2 types of data (POS + order management)
  • Target sends only POS data
  • Need to align on revenue recognition definitions (net revenue calculation)
  • Finance reconciliation: <10% difference is normal (accounting vs operational data)

3. Access & Self-Service

  • Snowflake AI Analyst coming (Uttam mentioned this)
  • Omni BI tool demo this week
  • Goal: Enable her to answer questions without needing working sessions

4. Process Alignment

  • Set expectations on iteration cycles
  • Balance between speed and quality
  • Define when working sessions are needed vs async updates

📝 Key Quotes from Shivani (Past 2 Weeks)

“Have you been able to join any Walmart and Target data? Like, have… like, if I were to say, what were sales across those two stores yesterday, would you be able to find it?”

“What are you actually seeing in, like, the delta between the Target and Walmart data right now?”

“If we’re gonna jam, then I… we need something to jam about. I’m not trying to be… I’m not trying to be difficult, I’m just, like, if we’re gonna jam and, like, do these sessions, then I need, like, a topic at hand”

“What is gonna make all of this work successful is if it lands for the, like, the people at LMNT to be able to answer basic questions. Everybody’s like, we have this retail data, but, like, what can I actually glean from it right now?”

“How is Target performing versus Walmart right now? Like, I don’t… I don’t know where I would find that, right?”

🔗 Related Documents

  • LMNT_Account_Transition_Prep.md - Overall account transition context
  • LMNT_Uttam_Shivani_Thought_Partner_Analysis.md - Understanding her working style with Uttam

Graph View