Hi @deek_kodem,
This is a classic AEP gotcha where batch success does not equal profile update success they are two separate pipeline stages and each can fail independently. Here are the four things to check in order.
First check Profile ingestion specifically not just batch status. Go to Monitoring, select your dataset, and switch the view from Dataset to Profile. A batch can show Success meaning the data landed in the data lake successfully but profile processing can still fail or be skipped. Look for profile-specific errors here, this is the most common root cause.
Second check your merge policy. Go to Customer Profiles, then Merge Policies. If the segment you are checking uses a dataset precedence merge policy and your new dataset is ranked lower than an older dataset with conflicting values, the older values win and the profile appears unchanged even though new data was ingested. Switch to timestamp ordered merge policy to confirm whether this is the issue.
Third check identity stitching. If ECID in your batch records does not exactly match the ECID already in the profile store, even a formatting difference like uppercase vs lowercase or extra whitespace, AEP creates a new profile fragment instead of merging with the existing one. Go to a specific profile in the UI, click Identity Graph, and see if you have duplicate fragments for the same person.
Fourth check segment evaluation mode. If your segment is using batch evaluation it only runs once per day on a schedule. If the scheduled run happened before your ingestion completed that person will not show updated membership until the next scheduled evaluation. Switch to streaming evaluation if near-real-time membership is required or manually trigger a segment job.
The most likely culprit based on your description is either profile monitoring showing errors that batch monitoring hides, or a merge policy conflict. Check those two first. Hope that helps!