Skip to main content
A workflow is an automation that reacts to your data and runs a sequence of actions. You design it as a flow of nodes — a trigger plus one or more actions — and the data platform executes it.
For the exhaustive, field-level reference — every action’s config, all strategies, credentials, interpolation, and the run/step model — see Workflows (reference).

​
Triggers

A workflow is triggered by an entity type — product, product variant, brand, or inventory item — optionally filtered by tags (include/exclude).

​
Actions

​
Strategies for “add items”

  • Quantity: fixed, fill to target, or score-driven (uses decision-vector recommendations).
  • Matching (rebalance): fixed or score-driven (matches surplus shops to deficit shops).
  • Discount (markdown): fixed, score-driven, or recommended (decision-layer discount with margin floor and caps).
Shop scope can be all, specific, or from the trigger; size scope can be all or below a stock threshold. Dates (deadlines, markdown dates) can be absolute, relative (days from now), or interpolated from earlier step outputs.

​
Runs & steps

Each execution is a workflow run (pending → running → completed / failed / cancelled), and each node produces a step record (pending, running, succeeded, failed, skipped, cancelled) with its resolved inputs, output, logs, and retries — so you can see exactly what happened. A workflow itself has a status: draft, active, paused, or archived. Only active workflows execute.

​
Working with workflows

1

Open Workflows

Go to Orchestration → Workflows.
2

Design the flow

Add a trigger (entity + tag filters) and chain actions in the full-screen builder.
3

Activate

Set the workflow to active so it runs when its trigger matches.
4

Monitor runs

Review run history and per-step details to verify outcomes.
Webhook credentials are stored securely and referenced by the Call-webhook action, so secrets never live in the workflow definition itself.