AEP failed batch analysis in one click [Lens]

ADOBE EXPERIENCE PLATFORM (AEP)

Pradeep Jaiswal

9/28/20265 min read

You open the run in Adobe Experience Platform. You get an error code, INGEST-1555–400, a one-line description, “Missing Required Field {field}”, and a count. You don’t get the field name filled in, the record that was missing it, the source file it came from, or the value it actually held. The data you need does exist. AEP keeps a copy of the rejected input, and when error diagnostics are on it also keeps a per-record error log. It sits behind a chain of export APIs: listing endpoints that nest folders inside folders, paginated responses, binary Parquet files, and multi-megabyte JSON Lines files. Reaching it by hand means Postman, a fresh bearer token, the right sandbox header, and about half an hour of copying `_links.self.href` values from one response into the next request. Analyze Failed Batch in Lens makes that chain one click.

Chrome extension. Link

What Is Lens?

Lens is a Chrome extension built for teams running Adobe Experience Platform. It works inside the AEP interface you already have open and uses your existing session: no extra login, no API credentials to configure, no data export step. Earlier articles in this series covered the Schema ERD Canvas, the Source Dataflow Canvas and the Approved Component Highlighter for CJA and Adobe Analytics. This one covers the feature AEP engineers tend to open the moment something breaks.

Three Ways In

1. The Analyze in Lens button

Open a failed batch on a dataset’s activity page, or a failed run on a dataflow’s activity page. A purple Analyze in Lens button appears next to the red Failed status in the right-hand panel. Click it and the Lens drawer opens straight to the analysis.

2. The row icon

On a dataflow Activity page with several failed runs, Lens adds a small orange magnifier next to every Failed status in the table. You don’t have to select a run to open its panel first; click the icon on the row you care about.

3. The search box
Someone pastes an id into a ticket or a chat. Type it into the search box at the top of the tab and press Search. The search is deliberately independent of the page you happen to be on.

Try Lens

The Header

The top of the tab tells you the batch status, the dataset and dataflow behind it, and how many records went in and how many failed, all in one glance. Whether you paste a batch id or a run id, you land on the same answer.

Batch Level Errors: What Actually Happened

Next comes every error Adobe attached to the batch and its run, each with its code and description, in one list. Run errors and batch errors are merged with duplicates removed, so you see the full story once. Typical entries:

- `CONNECTOR-2001–500`: processing failed after the data was copied
- `INGEST-1212–400`: the summary (“failed to ingest 2 rows”)
- `INGEST-1555–400`: a required field was missing
- `MAPPER-3707–199`: a source field in the mapping wasn’t found

This is useful on its own, but it only tells you what kind of failure happened. The sections below tell you which records and why.

Row Error File: See Exactly Which Records Failed and Why

Instead of a count and an error code, you get every failed record in a table, with the reason next to it.

- The exact error per record: code, failing column and message, so you know whether it is a mapping fix or a bad source value.
- The value that caused it: the record’s own source fields sit in the same row as the error.
- The source file it came from, so you know which upload to go back to.
- Fast to open, even on big batches: the first records appear in seconds, and you load the rest only when you need it.

If error diagnostics were off for the batch, Lens says so and tells you where to turn them on for the next run, instead of showing an empty table.

Batch File: The Input That Was Rejected

This is the original file that was rejected, before any mapping. It lets you check the data you actually sent against the error you were given, so you can confirm whether the problem is in the source file or in the mapping, and hand the exact file to whoever owns the source system. CSV, TSV, JSON and Parquet files all open as a readable table, with a dropdown when the batch had several files.

Working With the Records

Both sections share the same controls:

- Search filters the records as you type, across every column
- Pages of 10 records, up to 10,000 records per file
- Collapse chevrons on the errors list and both file sections, so you can focus on one at a time
- Export buttons named after the file they save: `csv`, `tsv`, `txt`, `json` or `parquet`.

Nothing is uploaded anywhere. The records are read from Adobe with your own session, shown in your browser, and discarded when you close the tab.

Who It’s For

Data engineers stop hand-walking export APIs. The question “which records failed and why” is answered in seconds, with the source value in the same row as the error.

Implementation consultants can diagnose a client’s broken ingestion live on a call, from the page the client already has open, without asking for API credentials.

Support and operations teams get a shareable record of what failed: search the rows, export the file, and attach it to the ticket.

Data owners can see whether a failure is a one-off bad file or a mapping problem affecting every run, and fix it at the source instead of re-running and hoping.

Practical Tips

- Turn on Error diagnostics for any dataflow you care about. Without it, Adobe keeps the rejected input but not the per-record reasons, and the Row Error file section has nothing to show.
- Act within 30 days: Adobe deletes diagnostic files after that.
- Start with conversion errors: They name the failing column and are usually a mapping fix.
- Use search on the Batch file to find the exact source row an error refers to, then fix it in the source system.

A red Failed badge is where most investigations begin. With Lens, it’s also where they can end: the errors, the failed records and the file they came from, one click from the page you’re already on.

Keep Exploring Lens

Analyze Failed Batch is one part of how Lens makes your Adobe Experience Platform setup visible. To see where the data comes from, read source-dataflow, which traces each source flow to the dataset it feeds. To see how your schemas relate, read schema ERD, which turns your schema metadata into an interactive ERD. And if your team reports in Customer Journey Analytics or Adobe Analytics, read Component highlighter, which shows how approved components are highlighted right inside the workspace.

Need Expert Help Fixing Adobe Experience Platform Ingestion Failures?

At Shiftlytic, we specialize in Adobe MarTech solutions, helping brands like yours find why batches fail, fix data quality issues at the source, and keep Adobe Experience Platform pipelines healthy. Whether you're stuck on recurring failed batches, unclear dataflow errors, XDM mapping and schema mismatches, or a source connector that keeps dropping records, our team of Adobe-certified experts is here to guide you.

Don't let failed ingestion leave gaps in your profiles and reports. Contact us today to schedule one free consultation and get your data flowing reliably, accurately, and ready for activation. Let's turn your data pipeline into a competitive advantage together.