Build Models for AI features (Auto-Target, Auto-Allocate etc) based on specific environments - Status changed to: Delivered

[#1]

Description - We have been trialling a number of Auto-Allocate/Target activities and we have unearthed that due to our geographical position we are unable to separate specific market/environment activities from the default environment when there are multiple environments present. The attempts to trial market specific activities results indicate that AI does not have an impact and the added reports (Pers. Insights/Important Attributes reports) not present. With the current documentation outlining that any activity that incorporates the default and another environment that it outlines the hits and conversions are 'Considered' but essentially requests are all served by the default environment.

Why is this feature important to you - It is important to be able to run the AI tools within Target that are not reliant on using and including the default environment which is a huge limitation for us. This enables market separation when launching a new AI delivered activity. 

How would you like the feature to work - When creating a new Auto-Target/Allocate/Personalisation activity having the option to determine which markets/environments for the activities models to be based on(Single or Multiple). Currently any test has to include the default, each of the activities should be dependant on the environment/market you want to run an activity on or on multi environment activities to select what environment is the decisioning source. 

Current Behaviour - We are currently using Auto-Allocate and Auto-Target, and whilst we have seen success in these tools we've observed and proven that the tools are not usable if you use within markets other than the default environment. For those users who work geographically and want to launch specific experiences for specific markets we are unable to without including the default environment, even then the data from the market is only 'Considered' as apart of the model not a decisioning element. 

[#2]

Thank you for your feedback!

In the documentation, here, it is clear that the models are built based on the traffic and conversion behavior recorded in the default environment only. After the models are built, if a hit occurs in another (non-default) environment, traffic is distributed according to the observed conversion behavior in the default environment. 

I understand that you want to have auto-allocate and auto-target build models off of traffic from any environment and we will keep that in mind for future iterations of the feature. For now, if you want the models to be built off traffic and conversion from another environment you can change the default environment. See instructions here. 

Thank you again for sharing your feedback! We love hearing what our customers would prioritize improving.

[#3]

Thanks for your reply!

What you have outlined is the opposite resolution to what we are requiring, having an environment dependency on auto-allocate/target activities rather than being built solely on the default environment. We have a wide range of geographical markets that need to be dependant from the default environment therefore the models will not be fully optimised for that specific market/environment.

Changing the environment is not an option for us as stated, we have a number of environments for each market due to other requirements of the business. So changing the environment would not be a resolution going forward as we'll have dependencies on the current default environment and constantly changing the default environment will not be good practice

I appreciate you considering this for a future iterations, but not sure I agree with it being 'Delivered'

@schristensenadobe 

[#4]

Thank you for the clarification! Knowing more helps us to understand why the current functionality doesn't fit your needs.