Where Share Moved

Category diagnosis

Rank by

Share movement waterfall

Opening share to closing share, decomposed by a fixed precedence ladder. Click any bar to filter the whole page to that driver.

Upstream Downstream External Unattributed

Page filter

How a lost unit gets exactly one owner

Deterministic, not modelled. Each rung repairs one condition and measures only what the rungs above it could not explain, so no unit is counted twice. The same ladder runs at every platform × category cell — which is why the cells add up to the roll-up exactly.

Δshare_driver (pt) = loss_units_driver × realised_price ÷ category_sales × 100

Upstream A PO existed and we did not serve it properly. Under-served against a confirmed PO, poor serviceability, or a lead time longer than the cover remaining. The fix is ours: ship more, ship faster, hold the commitment. Supply Planning · Logistics · Manufacturing

The axis Who owns the fix On a marketplace the platform raises the PO on us. So "no PO" is their decision, not our failure — which makes it a commercial conversation, not a supply one.

Downstream The platform never ordered, or never shelved what we sent. MOQ at that warehouse sits above their reorder trigger, or stock reached their DC and never made the dark store. The fix is a conversation with the platform. CAM · Key Account · Platform ops

The test. Does the fix start with a call to our supply team? → Upstream. Does it start with a call to the platform — an MOQ, an allocation, a listing? → Downstream. The point of the split is that each bucket lands on one desk with one conversation to have.

The honest limitation. The precedence order is an assumption, not a finding. Move Content above Price & Promo and the split changes. That is the cost of a number a KAM can argue with — and the reason we publish the ladder on the screen rather than burying it.

Activity trace — where the units stopped

One chain from the demand plan to the shopper. Each stage shows what survived it and what leaked, and which side of the handover the leak sits on. The seam is fill rate.

Upstream · before handover Downstream · after handover

What counts as a real availability problem

A snapshot says a SKU is out of stock today. That is noise. A problem is a SKU that keeps being out of stock at the same location — so the definition is persistence, and both halves of it are yours to set.

Unavailable in

or more of the last

crawls at that SKU × location.

Underneath the gate — what actually went wrong

The gate says who owns the fix. The reason says what happened. Internal reasons are ours to solve; account reasons sit at the platform end and need a different conversation.

Internal Account-side

Reason codes come from the live out-of-stock dashboard rather than from us — credit line, appointment, licence and capping are real account-side stops that our four gates had nowhere to put. A rupee carries one gate and one reason.

Root-cause tree

The same chain as a decision tree rather than a funnel — the shape the supply team already thinks in. Follow the failing branch down; every node carries the rupees behind it.

Raise a PO here — before the stockout

Everything else on this page is a post-mortem. This is the only list that is not: SKUs whose cover runs out before an order raised today could land.

Raise-by date = days of cover minus lead time. A negative number means the window has already closed and the stockout is now unavoidable — those rows are the ones to escalate rather than order.

Where the loss sits — pin-code level

Bubble size is rupees lost, colour is the side that owns the fix. Grouped by city tier, drillable to pin code.

Roll up by

Upstream-led Downstream-led

Down to the PO

Upstream is settled at PO level — ordered, confirmed, received, serviceability. Downstream is settled at MOQ level — what the CAM takes to the platform.

Which KPI is responsible

One row per driver, and the KPI that moved it. Click a row to open its sub-causes — each one is a join into an existing module, not a new engine.

Where it moved

Share Δ and its single attributed driver, at SKU level. Click a brand to expand.

Lens

Actions, sized to the share they win back

Impact density = recoverable share points ÷ weeks to impact. Numbers on the plot key to the queue beside it. Switch the ranking to see how differently the queue reads when it is ordered by money.