Applied outcome

SellerFlow

SellerFlow is the software implementation of the Decision Intelligence Framework developed through years of applied marketplace research. It observes the market continuously, establishes what each unit actually costs, and decides whether an action serves the objective of the business before that action is taken.

It is described here as the outcome of the research. This is not a product page.

SignalCompetitor offers, Buy Box, delivery
InterpretationBehaviour, position, intent
ConstraintCost, margin, inventory, velocity
DecisionAct, hold or withdraw
PriceThe output, not the objective

Why it was built

No available system decided anything.

The commercial tools available to a marketplace seller automate execution. They change prices quickly and reliably, and they do so without any model of the business behind the price.

Operating a catalogue through them made the gap concrete: the numbers moved, the decisions were still being made by hand, and the costs the system used to justify its moves were estimates that did not match the settlements.

SellerFlow was built to close that gap. It began as instrumentation — a way to record what competitors actually did and what each sale actually earned — and grew into the decision layer that record made possible.

Because it runs a live business, it is developed under real constraints: it must be accurate about money, resilient when a source of data fails, and explainable when it makes a decision an operator disagrees with.

What it does

Observation, economics, decision, execution.

LAYER 04 — EXECUTION Price submission · order fulfilment · multi-carrier shipping and tracking LAYER 03 — DECISION Strategy selection · competitor profiles · contest / hold / withdraw / recover records the reason for every decision LAYER 02 — ECONOMICS Settled fee reconciliation · cost of goods · shipping · true floor per unit one question per unit: what does this actually earn? LAYER 01 — OBSERVATION Continuous capture of buyer-visible offers · seller identity · delivery promise distributed workers, health monitoring, fallback data source
Figure 2. Reference architecture. Observation and economics feed the decision layer; no layer above may assume the state of the layer below.

Continuous market observation

Captures competing offers as a buyer sees them — seller, price, delivery promise, fulfilment channel — and maintains a behavioural record per competitor rather than a single market price.

Real unit economics

Reconciles settled marketplace fees, purchase costs, shipping and returns per unit, so that every pricing limit rests on what was actually charged.

The decision engine

Evaluates position, competitor behaviour, inventory, velocity and margin before choosing to act, hold or withdraw — and records the reason for the choice.

Operations

Order fulfilment, multi-carrier shipping and tracking, accounting and cost of goods, so that the decision layer works from the same figures as the business.

Relation to the research

The system is where the findings are tested.

Each study in the research index corresponds to a component of the platform. Competitor behaviour becomes competitor profiles; Buy Box probability becomes the estimate of what a position costs; fee reconciliation becomes the pricing floor; the decision framework becomes the engine itself.

The relationship runs in both directions. Operating the platform produces the observations that start the next study, and the record of changes made to it — problem, change, module, measured effect — is maintained as the primary evidence of how the work has developed.

Read the research index →

Enquiries

Technical enquiries welcome.

Available to discuss the architecture and the research behind it with journalists, analysts and fellow developers.

Contact