Start with the re-engagement question

Import the E-commerce scenario and separate the order, customer, and result assets needed for a campaign decision.

Question for this chapter

What data is needed to propose a customer's next action?

Why this matters now

Defining segment names first makes it easy to invent customers the data does not contain. Start by separating what order behavior, customer context, segment results, and operating metrics can tell you.

Try it

ecommerce.zip Download(32 KB)

In the left sidebar, select Collections. On the Collections page, choose the upper-right more (⋯) menu → Import (가져오기) and upload the ZIP. Select Start import after validation completes.

Success looks like this

The collection tree shows ecommerce_raw (E-Commerce Raw Data) and analytics (Analytics Data). analytics contains 100 raw_orders rows and 22 raw_customers rows. The product_manual knowledge asset belongs to ecommerce_raw.

Korean Portal showing source order and customer data with analytics assets in the analytics collection — The Portal capture is in Korean. Verify the source datasets and the assets that produce cleaned and analyzed outputs.

Interpret the result

Evidence	Question it answers	Asset to inspect
Orders	When did a customer buy what, and for how much?	`raw_orders`
Customers	Who are they, and what region and tier are they in?	`raw_customers`
Segments	What state follows from recency, frequency, and spend?	`customer_analytics`
Operating metrics	Where should campaign effort be concentrated?	`sales_overview` dashboard

Deep dive — assets imported by the scenario

Asset type	Count	Contents
Collections	2	`ecommerce_raw`, `analytics`
Datasets	4	raw orders and customers, cleaned orders, customer analytics
Codes	4	null cleanup, join, RFM, ontology materialization
Pipelines	3	order cleaning, customer segmentation, ontology materialization
Ontology	3 entities / 2 relations	customers, orders, products, purchase and containment relations
Knowledge / dashboard	1 each	`product_manual`, `sales_overview`

Next decision

You have separated the necessary evidence. Next, reduce 100 orders to usable rows and inspect the empty identifiers that remain after null cleanup.