To: Emma Rodriguez (Managing Director), David Chen (CFO), Marcus Kim (CIO)
From: Dr. Priya Sharma, Head of Data & Analytics
Now that funding is approved, this note gives my honest read on data readiness. I am confident the value is real, but the scoping numbers underestimate the data-quality and legacy-integration effort. I want that on the record before we commit to a timeline.
Where the data lives
| Source | System | State |
|---|---|---|
| Online sales & stock | Shopify Plus | Clean, API-accessible, near real-time |
| In-store sales | Legacy POS (~12 yrs) | Nightly batch export, inconsistent SKUs |
| Stock-on-hand by store | Legacy inventory system (~14 yrs) | Manual adjustments, ~7% variance vs counts |
| Supplier lead times | Spreadsheets / email | Unstructured, not systematised |
The core problem: siloed and inconsistent
Online and store data do not share a single product master. The same item can carry different SKUs across Shopify, the POS, and the inventory system. We estimate ~22% of SKUs need mapping or cleansing before they are usable for forecasting. A forecast is only as good as the stock-on-hand figure it starts from, and store-level accuracy currently sits around 93% against physical counts.
My recommendation: start with one category
Rather than forecasting every store and channel at once, I propose a single-category pilot. Homewares is a strong candidate (fewer style variants than fashion, more stable demand). We prove the data pipeline and the forecast quality on a contained footprint, then scale.
- Limits the blast radius of a bad forecast while we tune.
- Lets us measure real uplift before committing the rollout tranche.
- Forces the product-master cleanup on a manageable slice first.
Happy to walk through the SKU-mapping numbers with Marcus's team whenever suits.
Dr. Priya Sharma, Head of Data & Analytics