Skip to main content
An automation is a flow that runs itself: something happens to a store, Meridian checks whatever you want checked, and takes the action you configured. It’s how a Shopify app follows up on installs, churn, subscription changes and usage without anyone watching a dashboard. Automations are available for public apps.

​
The shape of a flow

Every flow is three kinds of step, connected on a canvas: Build it in the flow builder, or start from a template.

​
Active and inactive

Each automation is active or inactive as a whole. An inactive flow is never dispatched, so you can build one over several sittings and switch it on when it’s ready. Only active automations are considered when an event arrives. The list shows what you need to trust a flow: how many are active, runs over a chosen date range against the equal-length window before it, success rate with a failure count, and median and p95 run time. The shared range picker defaults to the last 30 days and uses UTC day boundaries. Search the list by name or description and filter it by status. An automation’s description shows under its name, cut to one line: hover it to read the whole text. To change the name or the description, open the automation, click More actions and choose Rename. A description holds up to 2,000 characters, and a counter beside the Description label shows how many you have used. The Name, Status, Last updated and Nodes headers sort the list: one click orders by that column, a second reverses it, a third returns to the default order. The sort survives search and filter changes and lives in the URL, so a sorted view can be shared. To jump to an automation from anywhere in Meridian, use global search.

​
What can start a flow

Twenty-three triggers, in five families: Each trigger carries its own variables, which conditions branch on and emails substitute.

​
What a run is

One event that matches an active automation queues one run of that automation. If three automations watch the same trigger, one event produces three runs, each with its own record. Runs are visible per automation with their status and failure reason. See Runs and troubleshooting. Before sending anything, you can test the flow against a real store or a synthetic one, and watch every step trace.

​
Allowances

Automation runs are metered per billing period, and so are the emails they send. When an app is at its monthly automation allowance, a new run is recorded as blocked rather than executing. Nothing is sent, and it isn’t retried. Both meters appear in Billing, where a pack can raise them. Slack messages have no meter of their own; only the run that posts them counts.
Last modified on September 30, 2026