Skip to main content

Migrating to the v5.4 Ecommerce package

The Ecommerce tracking package was introduced in Android and iOS v5.4.0. It provides 11 out-of-the-box event types to make it easier to thoroughly track user activity in an e-commerce store. A complete setup journey, including data modeling and visualization, is showcased in the Ecommerce Accelerator.

note

Migrating to the new ecommerce events is a breaking change for any relevant data models.

Replacing the deprecated eventsโ€‹

To update your existing tracking based on the now deprecated Ecommerce (iOS)/EcommerceTransaction (Android) and EcommerceItem/EcommerceTransactionItem event types, we recommend using TransactionEvent.

Old API:
let product = EcommerceItem(sku: "sku1", price: 100, quantity: 1)
.name("itemName")
.category("category")
.currency("GBP")
.orderId("id-123")

let event = Ecommerce(orderId: "id-123", totalValue: 350, items: [product])
.currency("GBP")
.taxValue(10)
.shipping(15)
.city("London")
.state("N/A")
.country("UK")
.affiliation("affiliation")

tracker.track(event)

New API:

let product = ProductEntity(
id: "sku1",
price: 100,
quantity: 1,
name: "itemName",
category: "category",
currency: "GBP"
)

let transaction = TransactionEntity(
transactionId: "id-123",
revenue: 350,
products: [product],
currency: "GBP",
tax: 10,
shipping: 15
)

let event = TransactionEvent(transaction: transaction, products: [product])

tracker.track(event)

These code snippets show only the direct API replacement. Check out the docs page or the Ecommerce Accelerator for information about expanding your tracking with the other new event types.

Note that address details and affiliation are not captured in TransactionEvent. To track these so that they are attached to the TransactionEvent, we suggest creating custom schemas and adding the data as an entity. Alternatively for addresses, you could use the CheckoutStepEvent which has address and postcode as optional parameters.

How the tracked events are differentโ€‹

Old eventsโ€‹

Using the old event types, an event is generated for each EcommerceTransaction event as well as each EcommerceTransactionItem attached. Therefore, tracking an EcommerceTransaction event with 3 items would result in 4 events being tracked. These events are not linked at all, unless the same orderId is provided to all the objects, adding potential challenges in data modeling.

In the tracked data, the events have event_name transaction or transaction_item. The data is sent within the legacy event properties tr_x and ti_x, e.g. tr_id for the EcommerceTransaction orderId property. All of these properties are listed here. Each one maps to a column in the warehouse table.

New eventsโ€‹

The new Ecommerce event types take advantage of the Snowplow entities/event context feature. Tracking TransactionEvent with 3 items results in a single event being tracked. The items, and the details of the transaction, are attached as entities. That single event would have one TransactionEntity, three ProductEntity, plus whichever out-of-the-box and/or custom entities are configured.

All of the new events are SelfDescribingEvent with event_name snowplow_ecommerce_action. None of the data is sent as legacy properties; it is all within entities, which end up in their own table columns.

Entities such as ProductEntity can be added to several of the new Ecommerce events, e.g. ProductViewEvent or AddToCartEvent. Reusing the same entities makes modeling easier and keeps the code cleaner.

Using the DBT data modelโ€‹

Since the structure of the events is so different, migrating to the new ecommerce events is a breaking change for any relevant data models.

We provide a dbt package that creates a set of derived tables including carts, transactions, and products from the raw ecommerce event data. Find the details of configuring this inside the Ecommerce Accelerator.

Was this page helpful?