Skip to main content
M
MahimaUp
Level 1
April 24, 2026
Solved

Handling Incorrect Data in Profile‑Enabled Datasets Without Impacting Existing Profiles

  • April 24, 2026
  • 1 reply
  • 33 views

We have an existing dataset that is enabled for customer profiles. Recently, some incorrect or undesirable data has been ingested into this dataset. I am looking for expert guidance on whether it is possible to remove or correct only the affected records without impacting the already established customer profiles or downstream profile calculations.

Specifically, I would like to understand the recommended approach or best practices for safely remediating bad data in a profile‑enabled dataset, while preserving profile integrity and minimizing any disruption to current profiles or dependent use cases.

  1. if we delete the that particular batch from dataset , does those are removed from profile store ?
  2. after correcting the data , if we re-ingest does the event data is overwritten with the latest data , 

Suggest me how can i resolve this 

Best answer by hegi90

Hi

So there are two important things. If you add Batch Data, this is added to the Data Lake (Update), and to the Profile Store (Upsert). If you delete the batch, only Data Lake data is deleted. It doesn’t remove data from the identity graph or profile store. If you re-ingest, profile store will be overwritten with the newest values (upsert).

You can also use Data hygiene features, to delete data, but this will delete the whole e.g. profile and identity graph, so be careful with that. So I would recommend:
Delete Batch and Re-Ingest clean data (delete datalake batch data and upsert on profile)

1 reply

H
hegi90Accepted solution
Level 3
April 28, 2026

Hi

So there are two important things. If you add Batch Data, this is added to the Data Lake (Update), and to the Profile Store (Upsert). If you delete the batch, only Data Lake data is deleted. It doesn’t remove data from the identity graph or profile store. If you re-ingest, profile store will be overwritten with the newest values (upsert).

You can also use Data hygiene features, to delete data, but this will delete the whole e.g. profile and identity graph, so be careful with that. So I would recommend:
Delete Batch and Re-Ingest clean data (delete datalake batch data and upsert on profile)

    Terms & ConditionsAccessibility statement

    Loading footer...