Offline Data Upload

[girishk10214964]

Hi Team,

we have a usecase where there is a content website where there are multiple articles. We want to upload offline data of the users who have read which articles and when( basically maintaining the history of the articles read) and accordingly will personalize those users. There will be multiple such rows corresponding to same user id. So how will that data will be utilized/ analyzed in AAM?

Example: (same userid , multiple rows but same keys)

Userid= 1, Article= Food , Date= 5/01/2017

Userid= 1, Article= Style , Date= 4/05/2018

Userid= 1, Article= Fashion , Date= 3/04/2019

How will AAM take up this file with same userid have multiple rows? Will this user id will be considered as the one who has read food, style and fashion article corresponding to the dates?

Elaborating more: As we have a limit of 200 lines we can process for each user ID sent in the inbound data file.

<user ID1><TAB><trait ID>,<trait ID>,<trait ID>

<user ID1><TAB><trait ID>,<trait ID>,<trait ID>

<user ID1><TAB><trait ID>,<trait ID>,<trait ID>

I want to understand how about same key but different values are going to work and what data will come in AAM corresponding to <user ID1>

<user ID1><TAB><Article= Food>,<Date= 5/01/2017>

<user ID1><TAB><Article= Style >,< Date=4/05/2018>

<user ID1><TAB><Article= Fashion>,<Date=3/04/2019>

[ryhamilt]

Hello!  As you mentioned in reference to the AAM product documentation (Inbound Data File Contents: Syntax, Invalid Characters, Variables, and Examples)​, Audience Manager will import the first 200 lines for a given user ID.  This is true whether you are passing in trait IDs, or whether they are different key value pairs.

As the data files are processed, Audience Manager will process the contents for each line.  As such, using your example of a user with the article of "Food", "Style", and "Fashion", AAM will realize the corresponding traits for each signal.  Once we have processed 200 lines for a given user, we will begin to discard the rest.

You also asked a question about whether the realizations will correspond to the dates provided.  If the intention is to notify AAM that the realization took place on that date, the system is unable to make that kind of association.  If, however, that is a metric that you are tracking separately for your own trait creation, you can certainly do so.  The only associations that Audience Manager makes regarding a particular set of signals for a given day, is based on when the file is processed.

[franciscor54562]

ryhamilt​true
Does it means that we can tie a product and its date in a trait?

ie:
Trait [ Article= Food AND Date= 5/01/2017 ] trait qualify

Trait [Article= Food AND Date=4/05/2018] Trait doesn't qualify
Question1: Am I right?

On the other hand,  if a segment had been created like this, would this user qualify?:

Segment:

Trait1 ( Article = Food )
           AND

Trait2 (Date = 4/05/2018)

As you can see in the records this product wasn't bought in this date, but the segment in this case is qualifying, because the user had qualified for both traits.

Question2: Am I right?

Question3: I know is not possible add recency or frequency to a onboarded trait. So, how could I create a segment to describe a person who bought a fashion article 30 days ago?

Thank you!

[Ankita_Sodhi · Accepted solution]

@franciscor35820985

Thank you for reaching out to us!

Please find the below inputs from my side:

Question1 - You can tie article and date but the date here will be considered as a string only. AAM doesn't identify date as an DateTime object and all the values are treated as string only. If you are storing date as a string at your end and would like to send this information to AAM then you can do so post which you can utilize this information to create traits/segments as mentioned in your example.

Question2 - User should qualify for segment with trati1(Article = food) AND trait2(date = 4/05/2018), provided this user appears online post the offline data upload and you are either sending this information in an offline file as signals.

Question3 - You might want to think about adding one rule-based trait which can capture some information for this user apart from the offline data and then apply recency/frequency settings on this rule-based trait.

I hope this helps, please let me know for any addition question/concern which I can help you with.

[franciscor54562]

Thank you for you answer! It helped me