Invalid profile qualified in segmentation

[shubham10]

Hi All,

In my source file I noticed conflicting values for a specific attribute  for the same identity and ingested data in the customer profile dataset as shown below:

Customer ID	Gender	LastUpdated
ID1	Female	2025-01-13 12:00:00
ID1	Male	2025-01-13 12:00:00

After fragmentation, in profile store I can see the value updated as "Male" (see below screenshot). I've verified using API and it gives me the same result

However, when I created an audience it qualified for "Female" only. I have cross checked the merge policy and its selected as default timebased (see below screenshot)

In my sandbox, I used only default timebased merge policy only.

Please help me to understand is there anything that I missed something or how segmentation works?

[DavidRoss91]

Hi @shubham10 

Trying to understand better here.. Did you ingest a gender value of "Male" and "Female" for the same identity value, at the same timestamp? I know that this can potentially cause issues when ingested at the exact same time.

[shubham10]

Hi @DavidRoss91 yes, its for same identity and at the same timestamp.

What is the issue if the timestamp is exactly is same?

[DavidRoss91]

@shubham10 

For that specific profile's identity graph, are there more than 50 identities associated with it by chance?

[shubham10]

No @DavidRoss91 only 2 identities are present and linked together one is customer_id and the other ecid.

[PratheepArunRaj]

Dear @shubham10 ,

Is the attribute a part of the batch source/dataset? Because data ingested using a batch-based source will be evaluated nightly, even if it qualifies for streaming segmentation.

Thank You, Pratheep Arun Raj B (Arun) | NextRow Digital | Terryn Winter Analytics

[shubham10]

Hi @PratheepArunRaj yes its a part of batch data and I ingested this on 13th Jan 2025. Even after 15 days I can see in profile store its "male" but the profile is qualified for female in segmentation.

[nnakirikanti]

@shubham10 Can you share your segmentation screenshot? 

[shubham10]

@nnakirikanti here it is

[PratheepArunRaj]

No Clue @shubham10 ,

The only possibility is that the attribute values are updated at the same exact time, and there might be some issues with segment qualification. Can you once again update the Gender value to 'MALE' today and wait for a day to see the output? I would do it as a last resort before reaching out to Adobe Customer Care.

Thank You, Pratheep Arun Raj B (Arun) | NextRow Digital | Terryn Winter Analytics

[nnakirikanti]

@shubham10 Looks all good to me, please do follow the next steps as @PratheepArunRaj provided.

~cheers,

NN.

[shubham10]

Sure, thank you everyone.

[jayakrishnaaparthasarathy]

Hi @shubham10  - I would suggest to delete dataset and try to ingest one more time with new dataset. However, create a schema with specific gender field with enum values and ingest the data again and wait for the batch to complete its daily job once you create the audience. Not sure this will work, but I would try this and see how does the platform react. 

Thank you,
Jayakrishnaa P.

[shubham10]

Hi @jayakrishnaaparthasarathy currently there are multiple campaigns are running so I can't delete this now. But I'll try this one later. Thank you.

[TylerKrause]

@shubham10 Have you tried rebuilding the audience? Obviously not ideal, but we have unfortunately seen issues similar to this in the past. We opened up a support ticket to get it resolved, it required an Adobe back end refresh. 

I would additionally set up an additional audience for that customer id with an experience for male as well. If the profiles qualifies for both, that is a clear issue that would need to get resolved by the Adobe support team.

[shubham10]

Yes @TylerKrause , I created a test segment and can see the same result. I'll also create a Adobe ticket to understand why this type of issues occur frequently now.

[kautuk_sahni]

@shubham10 Did you find the suggestions helpful? Please let us know if you need more information. If a response worked, kindly mark it as correct for posterity; alternatively, if you found a solution yourself, we’d appreciate it if you could share it with the community. Thank you!