Expand my Community achievements bar.

Join us for the next Community Q&A Coffee Break on Tuesday April 23, 2024 with Eric Matisoff, Principal Evangelist, Analytics & Data Science, who will join us to discuss all the big news and announcements from Summit 2024!

Unifying Insights: The Power of Metric Deduplication in Adobe Customer Journey Analytics

Avatar

Employee

8/30/23

Unifying Insights: The Power of Metric Deduplication in Adobe Customer Journey Analytics

 

When leveraging Customer Journey Analytics (CJA) for analysis/reporting, it’s important to not inflate metrics that should be counting unique values.  CJA has a Metric Deduplication feature that you can use to ensure accurate representation of these metrics. This blog post will walk you through how to configure Metric Deduplication within CJA Data Views for some common use cases.


Metric Deduplication:
Metric Deduplication is a powerful feature that provides you with the ability to configure a specific metric in a manner that allows it to track and tally distinct values while eliminating any redundant occurrences. This functionality proves invaluable in assessing the individuality of events recorded and gaining a precise comprehension of their occurrence frequency.

When applied within the context of the CJA (Customer Journey Analytics) platform, Metric Deduplication offers a range of nuanced measurement options. You can opt to evaluate the metric based on either the first instance it is encountered or the last instance it appears. Moreover, this evaluation can be tailored to either the people-level, which assesses unique individuals, or the user session-level, which considers distinct interactions or engagements. This sophisticated level of customization ensures that the metric analysis aligns precisely with your analytical needs and objectives, enabling a more nuanced and accurate understanding of the data.

Agastya304415685kgd_0-1693393188105.png

Use Cases:
Metric Deduplication helps you to understand first and last instances of an event.

  • It helps you remove multiple Instances of orders with respect to same order ID.
  • It helps you remove multiple Instances of application submission with respect to same application ID.
  • It helps you remove multiple instances of user subscription Instances.
  • It helps you remove multiple campaign instances within the same session.
  • It helps you report on first-time purchasers within a specific timeframe.
  • It helps you keep the last instance of a form submission from a user.


In this blog we will focus on two common use cases as examples.


Example Use Case #1: Multiple instances of orders with respect to same order ID
If you are seeing multiple instances of same order ID in your reporting, follow the below steps to remove the repeated instances for the same order ID.

NOTE:  If you are familiar with Adobe Analytics' e-commerce purchase event and have implemented the purchase ID to remove duplicate orders, this feature in CJA helps you eliminate duplicates with more options.

Step 1: Select the respective Data View and click on Components Tab.
Step 2: Search for the purchase event [In my case it is commerce.purchases.value]
Step 3: Click on the event under Metric section and select Metric Deduplication. Provide the below details.

Deduplication Scope: - Person
We have people-level and session-level scope. We are selecting the person-level scope because we are analyzing orders of a user. If the Instance is scoped at the people-level, it is calculated based on the overall user journey with the application, not at just the reporting window that’s selected. If we have selected session-level It only consider the duplicate data only from the same session of the user, if the user has a duplicate event in a different session the data will not filter from the reporting.

Deduplication ID (optional): - Purchase ID
In this use case we have selected the Purchase ID because it is the unique Identifier for the order in the above example. If Deduplication ID is empty, it will consider the People dimension at person-level scope and Session ID at session-level scope.

Value to Keep: - Keep first Instance

We have selected to 'Keep first instance' of the Purchase ID. In our use case, we aim to filter out duplicated data with identical Purchase IDs. As a result, we have opted to preserve the data that was initially ingested with that specific Purchase ID. If we have selected 'Keep last instance' of the Purchase ID, it will override the existing data with respect to Purchase ID with last ingested data if the data is already ingested with that Purchase ID.

Agastya304415685kgd_1-1693393866558.png

Step 4: Click on Save and Continue and save and finish.

Now, if you look at the reporting in your workspace, you will no longer see duplicate events with respect to the Order ID.

Agastya304415685kgd_2-1693393906987.png

In the duplicates-freeform table, we can see the purchase ID populated 5 times. However, since Metric Deduplication was applied to commerce.purchases.value, the unique event freeform table shows only 1 instance of the purchase ID displayed, indicating that duplicates have been effectively removed.

Example Use Case #2: I need to understand the first purchase order of users within a specific time-window
If want to know the first purchases of the users for specific time-window. Follow the below steps to get the first order instance of the users.

Step 1: Select the Respective Data View and click on the Components Tab.
Step 2: Search for the purchase event [In my case, it is commerce.purchases.value]
Step 3: Click on the event under the Metric section and click on ‘Duplicate’.

Agastya304415685kgd_3-1693393944623.png

Step 4: Select the duplicate created and rename it to “First Order”. Then, select Metric Deduplication and provide the below details.

Deduplication Scope: - Person
We have person-level and session-level scope; we are selecting person-level scope because we are considering the first order of a userIf the Instance is scoped at the people-level, it is calculated based on the overall user journey with the application, not at just the reporting window that’s selected.

Deduplication ID (optional): - User ID
We have selected User ID because it is the unique Identifier for the order in the above example. You can select a unique identifier from the respective event for which you want to remove duplicates. If it is empty, it will consider the People dimension at the person-level scope and Session-ID at session-level scope.

Value to Keep: - Keep first Instance
We have selected ‘Keep first Instance of the Order ID’ because it represents the first order placed by a user.

Agastya304415685kgd_4-1693393998536.png

Step 5: Click on Save and Continue and save and finish.


Now, if you look at the reporting, we can analyze the number of first orders in this Year, quarter, specific window, or in comparison with respect to time.

Agastya304415685kgd_5-1693394108249.png

User 62267…0268 made 3 orders this month, with the first order also being made in this month. If the user makes 2 more orders next month, they will no longer be considered a first-time order for that user. Now, I can personalize the website for these users, providing a specific first-order experience feedback form popup.

Conclusion: The Metric Deduplication feature in CJA helps businesses report and understand events counts more accurately.

The Metric Deduplication feature in CJA helps businesses count events more accurately. It removes duplicates, so you don't count the same thing multiples. This makes your reports and understanding of events much clearer and reliable, helping you make better decisions based on accurate information. By getting rid of repeated data, Metric Deduplication ensures that your event counts are spot-on and not inflated. This straightforward process enhances the reliability of your insights, giving you a solid foundation for making smart business choices.

Reference links:

https://experienceleague.adobe.com/docs/analytics-platform/using/cja-landing.html

https://experienceleague.adobe.com/docs/analytics-platform/using/cja-dataviews/component-settings/me...

1 Comment