Best practice on XDM schema for web and mobile app

[RaymondCh4]

Hi there

Currently we have implemented Web SDK on our websites; we customized slightly on the ExperienceEvent data schema but pretty much leverage most of it for our integration. 

As we are now progressing to apply Mobile SDK on our mobile apps. Assuming we have very similar features between our website and mobile app, I wonder is there any best practices regarding the XDM schema for the mobile app? In particular:

• Shall we use the exact same schema and data set for both the website and mobile app? 
• If yes / no, what's the pros and cons of doing so? 

Appreciate your insights in advance!

Thanks,

[Harveer_SinghGi1]

Hi @RaymondCh4 ,

It can depend on lot of things considering how you plan to use the data in current and future state of your AEP implementation. While you can use the same schema and dataset for both web and mobile their could be other things to consider which may suggest to separate these two data sources,

• For dataset specific controls like events TTL or deciding which platform to ingest into a CJA connection you'll have more flexibility with separate datasets
• If you are using (or planning to use) this data in profiles you will have more control in terms of merge policies for attributes with conflicting values
• Even if you have the same features in both platforms there are some attributes which are platform specific, like lifecycle data from apps. Keeping both platforms separate will give you more cleaner records with only respective fields
• Another thing to consider could be the ease of understanding and auditing/troubleshooting with different datasets

In the end how you plan you data models is going to depend upon how you plan to ingest and consume the data and that varies based on use cases that you want to serve.

Cheers!

[RaymondCh4]

Thank you for the comprehensive response!! Really appreciate it!

[Jennifer_Kunz]

Beyond the great points made by @Harveer_SinghGi1, I think it's mostly a matter of preference. Personally, if the requirements for both are generally the same, I think it'd be easier to maintain a single schema, though perhaps different datastreams and/or datasets.