Placing a decision layer between retail operations and merchandising decisions

← Architectures

Reference structure · Retail · Technology × Organisations

One retail architecture in which operational systems stay authoritative and a thin layer above them assembles the reading that decisions are made from.

This is one possible structure, not a target state and not a record of a system observed in a specific environment. It is useful while its assumptions hold, and it says where a different arrangement would fit better.

The structural question

Should merchandising decisions read directly from operational systems, or from a layer that reconciles them first?

Components

Each part is here for a reason: what it owns, and what it depends on.

• Operational systems (ERP, OMS, WMS, commerce platform)

  Owns: Source of record for inventory, orders, cost and fulfilment state.

  Depends on: Depend on nothing above them; the decision layer must not become load-bearing for operations.

• Reconciliation layer

  Owns: Source of record for the reconciled reading — not for any underlying entity.

  Depends on: Depends on definitions of margin, availability and returns being agreed rather than inferred.

• Channel and media data

  Owns: Their own performance records.

  Depends on: Depends on identifiers matching catalogue keys; usually the weakest join in retail.

• Decision surface

  Owns: The decisions taken and the state they were taken on.

  Depends on: Depends on the reconciled reading being timestamped so staleness is visible.

Ownership and source of record

ERP or OMS remains source of record for inventory truth; the layer above owns only the reconciled reading. Whoever can change the margin definition changes every decision made from it, so that rule belongs with finance rather than with the reporting team.

Movement through the structure

1. 01

   Operational systems

2. 02

   Reconciliation layer

3. 03

   Reconciled reading

4. 04

   Decision surface

5. 05

   Read-out

Most of the difficulty is definitional, not architectural: what counts as available stock, and whether margin includes returns and fulfilment cost.

Dependencies

The question is whether the dependency matches what the organisation needs.

Intentional coupling

Catalogue identity and channel data must move together, or performance cannot be attributed to a product at all.

Costly coupling

Letting the decision layer write back into operational systems couples analytical logic to fulfilment reliability.

What is hard to reverse

Agreed definitions are the durable asset and travel between tools. Dashboard and warehouse vendors are replaceable; a decision layer that has silently become the source of record for inventory is not.

When something fails

If the reconciliation layer fails, operations continue and decisions run on the previous reading — acceptable for assortment, less acceptable for pricing during a promotion. Staleness must be shown, not assumed.

Assumptions

A reference structure is only reusable when its assumptions travel with it.

What must hold

If these are not true in a given environment, the structure is the wrong starting point.

• More than one channel or system holds a partial view of the same product.
• Daily or intra-day latency is acceptable for most merchandising decisions.
• Definitions of margin and availability can be agreed across functions.
• Catalogue identifiers are consistent enough to join, or can be made so.

Trade-offs

Each improves something and costs something. Neither side is universally preferable.

• A reconciled reading against the effort of maintaining definitions as the business changes.
• Centralised numbers against local autonomy in how teams read their own channel.
• Current state against a smaller failure surface — not every decision needs real-time data.

What this makes possible

Assortment, pricing and creative decisions can be argued from one reading, which relocates the argument to the assumption rather than the export.

What it constrains

It makes team-local definitions harder to keep, and adds a component whose failure is felt in decision-making rather than in operations.

How you would know whether it helped

Whether decisions cite the same numbers, and whether stock-out and over-discount patterns change in a way that can be attributed to earlier visibility.

What informed this

Platform constraints in ERP, OMS and commerce tooling, technical reasoning about latency requirements, and observed situations in multi-channel retail environments.

Another structure may fit when

Another structure may fit when the ERP or OMS reporting layer is already sufficient — adding a reconciliation layer then duplicates definitions rather than settling them.

Where this stops fitting

Stops fitting single-channel operations, environments where one system already holds a sufficient reading, or organisations without an owner for the definitions.

Related

Related reading

Industry

• Retail
• D2C
• Consumer Brands

Knowledge · Playbook

• Turning Weekly Retail Signals Into Decisions Someone Owns
• Operating a Marketplace That Keeps Moving Underneath You

Evidence · Observed Situation

• When growth outpaced the organisation's ability to read itself
• When every team had a dashboard and no team had a decision

The situations linked above informed this representation. They do not establish that this structure is right for another environment — that remains a question about the specific organisation, its ownership and its constraints.

Architecture, as a question →Work with Cralgo →