Dry runs

A dry run does the work but doesn’t write the result anywhere. Toggle Dry run in the Start run modal.

Dry-run only applies to phases that produce a side effect — transform and load. Extract has nothing to dry-run, so the toggle is ignored when extract is the only phase selected.

Dry-run transform

• The mapping runs against the latest extracted data.
• The output is computed and shown on the run item, but not staged for load.
• Nothing changes in your queue of records waiting to push.

Useful for: previewing mapping changes against the whole catalog without affecting subsequent loads.

Dry-run load

• Each pending record is checked: would it create or update? With what payload? With what conflict resolution?
• Nothing is sent to Shopify.
• The run item logs every intended call so you can audit the destination’s diff before pulling the trigger.

The summary breaks the counts down per resource: how many records would be created, how many updated, and — for orders — how many would be rebuilt. A rebuild creates the corrected order first and then removes the original, which Shopify forces whenever an existing order’s products, quantities, prices, taxes or payment status change. It’s the one number here worth checking twice, and a dry run is the only place to see it before it happens. See Re-loading an order that already exists.

Useful for: cut-over rehearsals. Run a dry load the day before, review counts, then run the real load.

Updating an existing store

When the destination is the store you read from, Dry run is on by default — the safe first action against live data is always “show me what would change”. The summary reports updates rather than creates, because that kind of migration never creates a record.

Turning the toggle off asks you to type your shop domain before the run will start. See Update an existing Shopify store.

What dry runs cost

Dry runs still hit the source (if extract is included) and still spend time computing transforms. Cost-wise they’re identical to real runs minus the destination side. Use them freely during configuration; reach for transform-only without dry-run for the actual mapping iteration loop.