Repeat order¶

Repeat Order returns a list of customers who have made a repeat purchase during the chosen date range. For example, return all customers who made a repeat purchase within the previous 6 months:

Customers who own the base game and need to buy an expansion pack.

Important

Repeat Order identifies a repeat purchase by comparing purchases within the chosen date range to all purchases. For example, a customer who first purchased 2 years ago, and then purchased again last month would be returned by a relative date range “today - 1 month”. A customer who first purchased 10 years ago, and then purchased again last month would also be returned. As well as a customer who first purchased 2 months ago, and then purchased again last month.

After you specify a date range you may apply filters to associate repeat orders to specific products, brands, channels, and stores. For example, return all customers who made a repeat purchase of shoes from your ACME Footwear brand.

How this attribute works¶

Repeat Order represents a common approach people use when they build segments: find customers who have purchased more than one time within a time window, and then associate those customers to specific products and brands.

Repeat Order is a compound attribute, which means that it’s built from a combination of attributes that already exist in your data, and then appears as a single attribute that you can choose from the Segment Editor.

With this attribute, you can focus less on SQL and more on finding answers that align to your marketing goals and strategies. Purchase behavior attributes simplify the number of steps that are required to associate a list of customers to your products, stores, channels, and brands.

How does the SQL for Repeat Order work?

Repeat Order is built from standard columns that are output by Amperity. The following example returns a list of customers who have purchased at least one component of a game during the previous 3 years:

The SQL for Repeat Order works like this. It uses two common table expressions (CTEs) to find two lists: a list of first orders and a list of distinct orders.

To build the list of first orders, Repeat Order uses order datetime from the Unified Itemized Transactions table as its starting point:

SELECT
  amperity_id
  ,order_datetime AS first_order_datetime
  ,RANK() OVER (PARTITION BY amperity_id ORDER BY order_datetime, order_id) AS order_rank
FROM Unified_Itemized_Transactions

then uses the RANK() function to identify which of those orders were repeat orders, and then returns all items that match two conditions: order rank and a date range:

WHERE order_rank = 1
AND order_datetime < DATE_TRUNC('day', CURRENT_TIMESTAMP - interval '36' month)

This returns the first order for all customers who purchased during the previous 3 years.

Returns and cancellations are filtered out automatically, like this:

AND (is_return IS NULL OR (NOT is_return))
AND (is_cancellation IS NULL OR (NOT is_cancellation))

To build the list of distinct orders, Repeat Order uses order datetime from the Unified Itemized Transactions table to find all of the distinct orders that were made during the same time period:

SELECT DISTINCT
  uit.amperity_id
FROM (
  SELECT
    amperity_id
    ,order_datetime AS distinct_order_datetime
  FROM Unified_Itemized_Transactions
)

To identify which customers made a repeat purchase during this time window, the first order CTE is joined to the distinct order CTE:

INNER JOIN first_order_datetime
ON distinct_order_datetime

(Note that “first_order_datetime” and “distinct_order_datetime” names do not represent the real names of the CTEs that are joined; these names identify the two sources that are joined together as part of the steps within the SQL that identify your repeat customers.)

Returns and cancellations are then filtered out automatically, like this:

AND (
  is_return IS NULL
  OR (NOT is_return)
)

You may then filter this list of customers more by applying any of the product, purchase, and store filters. When you select these filters, they are added to the WHERE statement, like this:

AND product_category = 'game'
AND product_subcategory IN ('base', 'expansion', 'pack')
AND purchase_brand = 'Fearless Creatures'

The SQL for Repeat Order is more complex than what is described in the previous section. This is due to the way this attribute returns only a list of Amperity IDs, uses a series of common table expressions (CTEs), and takes advantage of workflows that Amperity does behind the scenes to pre-filter the product, purchase, and store attributes.

You can view the full SQL for Repeat Order from the Segment Editor. Start a new segment and add only this attribute (along with any required conditions and filter attributes), and then click the View SQL link at the top of the page.

Add to segments¶

To find all customers who made their first order within a specified date range, start with the Repeat Order attribute located under Purchase behaviors:

Use the repeat order attribute to find repeat orders by date range.

After the attribute appears in your segment, choose a date range:

After you specify a date range you may apply filters to associate these customers to specific products, brands, channels, and stores.

About relative dates¶

A relative date is determined at the time a segment is run, where today is the day on which the segment is run. For example: yesterday, 30 days ago, 14 days ago, or 1 year ago. The list of relative date values includes a series of common ranges, but you may also type in a more specific range, such as 2 months ago or 5 days ago.

Relative date values

Value	Description
Tomorrow	Starts at 12:00:00 AM of the day after the current day. For example, if the current day is “Thursday 01 / 12 / 2023” then tomorrow is “Friday 01 / 13 / 2023”.
Today	Starts at 12:00:00 AM (or at the current time) on the current day and continues for 24 hours. For example, if the current day is “Thursday 01 / 12 / 2023” then today is “Thursday 01 / 12 / 2023”.
Yesterday	Starts at 12:00:00 AM of the day before the current day. For example, if the current day is “Thursday 01 / 12 / 2023” then yesterday is “Wednesday 01 / 11 / 2023”.
N days ago	Starts at 12:00:00 AM of the day N days before the current day. For example, if the current day is “Thursday 01 / 12 / 2023”, then: 7 days ago is “Thursday 01 / 05 / 2023” 14 days ago is “Thursday 12 / 29 / 2022” 30 days ago is “Tuesday 12 / 13 / 2022” 60 days ago is “Sunday 11 / 13 / 2022” 90 days ago is “Friday 10 / 14 / 2022”
1 month ago	Starts at 12:00:00 AM of the same day of the month that is 1 month before the current month. For example, if the current day is “Thursday 01 / 12 / 2023” then 1 month ago is “Monday 12 / 12 / 2022”.
1 year ago	Starts at 12:00:00 AM of same day of the year that is 1 year before the current year. For example, if the current day is “Thursday 01 / 12 / 2023” then 1 year ago is “Wednesday 01 / 12 / 2022”.

Tip

You can compare segments by % of Purchasers, % of Revenue, or Revenue/Purchaser by changing the option in the Compare by: field.

Available conditions¶

The following table lists the conditions that are available to this attribute.

Note

Recommended conditions for this attribute are identified with “ More useful” and conditions with more limited use cases are identified with “ Less useful”.

Condition	Description
was after	More useful Returns a list of customers where repeat purchases were after the specified time window.
was before	More useful Returns a list of customers where repeat purchases were before the specified time window.
was between	More useful Returns a list of customers where repeat purchases were between two specified time windows.
was not between	Less useful Returns a list of customers where repeat purchases were not between two specified time windows.
was not on	Less useful Returns a list of customers where repeat purchases were not on the specified date.
was on	Returns a list of customers where repeat purchases were on the specified date.

Filter attributes¶

A filter attribute is a standard column that is output by Amperity and is available from the Unified Itemized Transactions table. When a filter attribute is associated with a purchase behavior attribute, you may use them to filter the results by specific items in your product catalog, such as by brand, by channel, by store, or by specific details about the items in your product catalog, such as color, SKU, and so on. The list of filter attributes that will be available for product catalogs depends on their availability within your Unified Itemized Transactions table.