Sunday, April 22, 2018

Key Adobe Audience Manager Metrics and Ratios

Each reporting system has its own methodology for calculating metrics and a Data Management Platform is no different. When I first learnt Audience Manager, one of concepts that took me a while to understand was how Audience Manager calculates its metrics and this post is intended to cover how it's done.

To begin, it's important to understand that all key metric calculation in Audience Manger is based off devices and cookies so AAM doesn't taken into account raw hits into its calculation. Below is a summary of how each of these are calculated taking into consideration the following visuals.

Trait Metrics

AAM traits are the most granular data points or signals that are populated in Audience Manager. They represent unique visitors that are calculated based off cookies or devices. Please note that by default, a trait may count the same person multiple times if that person qualified for a segment across multiple devices.

Unique Trait Realization: This is the total number of visitors that qualified for a trait on the specified time frame (1 day, 7 days, lifetime etc). This is the typically the preferred metric for gauging active visitors. An example of a trait can be visitors to a particular URL or visitors who completed a purchase. In the screenshot above, 18,981 visitors qualified for this trait during the last day and 2,791,994 visitors have qualified for this trait since it was created. 

Total Trait Population: This is the running total number of visitors that are part of this trait since creation based on the specified time frame. In this example, a total 2,539,320 visitors were part of this trait as of yesterday and 2,590,249 visitors were part of this trait in the last 7 days. The reason why the lifetime count of visitors (2,775,871) for this metric is less than that of Unique Trait Realization (2,791,994), is because this metric starts getting calculated after a gap of 24 hours compared to UTR. Please note that the TTP gets refreshed every 120 (expiration) so any user who is seen after 120 days is counted again in the TTP calculation.

Segment Metrics
Segments are built from a combination of individual traits and are shared as audiences to outgoing destinations for activation. If multiple traits mapped to a single segment have the same cookie or device, they are deduplicated at the segment level. Segments can either contain multiple traits or can contain an overlap between two separate traits among other conditional logic.

Real-time Segment Population: This the total number of users who qualified for a trait(s) based on the specified time frame. Similar to UTR, this number will change everyday based on how many users qualified for this segment. In this example above, 54,377 users qualified for this segment during the last day.

Total Segment Population: This is the overall segment population which were part of this segment based on the specified time frame. In this example, there were 4,633,044 unique visitors that were captured across multiple traits which is why the overall volume of segments is more than that of traits.

Destination Metrics & Ratios

Destinations allow segments to be mapped and shared with various outbound partners or DSPs for activated. The following metrics and ratios fluctuate everyday based on how many users have a match between Audience Manager and the DSP.

For this example, I've used an older screenshot with dummy data as I didn't want to use an example from an actual client. Please note that the following screenshot doesn't reflect what the UI looks currently and some metric names will not match entirely so my apologies.

Addressable Audience Match Rate: This is the percentage of visitors that have a device match/sync between the customer's Audience Manager instance and 3rd party partner to be activated on. It is the calculated as ratio of (Customer Addressable Audience [676,173,551]/Customer Total Audience [1,428,072,915]). The higher the match rate, the more the visitors that can be targeted in the DSP and in this example, it's 47%. To increase match rates between AAM and the DSP, the Adobe consultant turns on an ID sync in the backend. Some causes of low match rate are explained here.

Customer Addressable Audience (Devices): This is the total number of visitors that have a device match/sync between the customer's Audience Manager instance and 3rd party partner. The Customer Addressable Audience count shown in the screenshot is 676,173,551 visitors for the last day.

Customer Total Audience: This is the total number of devices that are active in the customer's Audience Manager instance. In this example, the total active devices are 1,428,072,915 which will fluctuate daily.

Audience Manager's Addressable Audience (Lifetime): This is the total number of devices that are active across all Audience Manager customers. Please note that this number will always be more than Customer Total Audience for a particular client account unlike the screenshot above which contains dummy data.

Segment Addressable Audience (Devices): The total number of users who have an active ID sync  for the specific AAM instance localized for a mapped segment. Unlike in the screenshot, this number will vary by each mapped segment which is 123,456,789 for the example.

Segment Match Rate: This is the percentage of Segment Addressable Audience/Total Segment Population. This match rate is more reflective of how many users can be targeted than the Addressable Audience Match Rate.

Profile Merge Rule Metrics & Ratios
Profile Merge Rules or PMRs uniquely identify a user across browsers or devices that allows marketers to deliver a consistent message to their customers. This is made possible by capturing the user's hashed authenticated ID across devices.

Total Devices: T
his is the total number of devices where a hashed authenticated user ID was captured in the last 60 days. In this example, the total number of devices is 3.69 Million.

Total Person: This is the number of unique users where a hashed authenticated user ID was captured across multiple devices. In this example, the number of unique users is 3.32 Million which is less than the total number of devices.

Average devices per Person: This is the ratio of Total Devices/Total People. In this example, an average user logs in across 1.1 devices which rounds up to a single device. Marketers can use this ratio to determine the total number of devices a user logs into.

Active People: This is the number of unique users where a hashed authenticated user ID was captured across multiple devices. It's the same metric as Total Person which is 3.32 Million.

Cross Device: The total number of Cross Device IDs stored for the selected Authenticated Profile since creation. It's basically the total devices in an Audience Manager Instance. In this example, it's 6,711,719. FYI, these are users captured in the last 120 days which is why they're higher than Total Devices.

Active People %: It is a ratio of Active People divided by Cross Device. In this example, it's 50%.

Some of these metrics are already explained in the official Adobe documentation in different places but I've attempted to structure and put them in one post based on my own understanding.


TestBlog said...

Nicely written and organized several bits of information in a central place. Awesome!

Sainath Revankar said...

Thanks for sharing valuable information. When you get a chance can you share some insights on Profile Merge Rule? As to when to appropriately choose Authenticate profile and Device profiles? Thanks for all your blogs so far.

Rohan Kapoor said...

@Sainath, the main difference between the current device (anonymous) and authenticated device (logged in) is that you can either target anonymous users (typically a bigger set of users) who have not logged in vs. authenticated users who have signed in assuming they have an id sync with the ID service. With auth profile, you can better personalize the users and show them tailored content based on their demographic (and web/app behavioral) data whereas with anon profile, you can show users ads based on their web/app behavioral data.

Sainath Revankar said...

This helps. Thank you, @Rohan.

Unknown said...

Nice blog with a lot of difficult to understand terms explained in a nice fashion!!. Great Job. Way more to go.

Rohan Kapoor said...

Thank you, for the comment!