Practical guide
Batch conversion needs per-file control
A batch is not one document multiplied by twenty. Real collections mix clean text PDFs, scans, tables, password-protected files, duplicate exports, and outliers with unusual page counts. DocKernel's batch panel accepts multiple PDF files for an authenticated job, shows aggregate progress, keeps item-level status, supports retries for eligible failed jobs, and provides downloads for completed items. That visibility matters because one bad file should not hide the results of the rest of the queue.
The current browser client protects the queue with explicit limits: up to 20 selected PDFs, up to 50 MB for an individual batch file, and up to 250 MB total selection size. Backend entitlements can apply additional limits. Before a large migration, group documents by type, remove obvious duplicates, run a representative sample, and define how names, folders, assets, and failures will be handled. Batch throughput is useful only when the output remains attributable to the correct source.
When batch PDF to Markdown saves real work
- A documentation migration contains many related manuals or release archives.
- A research team receives a recurring folder of reports with similar structure.
- An ingestion pipeline needs separate status and downloads for every source file.
- You can define sample-based review rules and an operator process for failed or unusual documents.
Workflow
Run a controlled batch conversion
- 01
Prepare the collection
Remove duplicates, split unsupported archives, normalize filenames, identify confidential files, and separate likely scans from normal text PDFs.
- 02
Select and submit
Choose up to the visible client limits, sign in, preserve the desired page and line-break options, and create the batch.
- 03
Monitor each item
Watch completed and failed counts, inspect item status, and retry only jobs that the service marks as eligible.
- 04
Download and quality-check
Save outputs with stable names, open a sample from each document type, compare source pages, and quarantine anomalies before the next system imports them.
Conversion example
Example: a mixed documentation batch
The manifest makes the outcome explicit. Completed files continue through the workflow, while a scanned file can be rerouted to OCR and an encrypted file returns to the owner instead of blocking the whole batch.
Source PDF content
Batch: onboarding-archive
- employee-handbook.pdf
- benefits-scan.pdf
- office-map.pdf
- encrypted-policy.pdfMarkdown output
employee-handbook.md completed
benefits-scan.pdf needs OCR review
office-map.md completed with visual page
encrypted-policy.pdf permanent input error
Summary: 2 accepted, 1 reroute, 1 owner actionQuality benchmark
Batch quality and recovery benchmark
Measure a batch by completeness, traceability, and recovery behavior. Average speed alone can conceal missing outputs or repeated failures.
| Document profile | Recommended route | Pass criteria |
|---|---|---|
| Homogeneous text manuals | One sampled batch with standard options | All items reach a terminal state; sampled headings and page markers match their sources. |
| Mixed text and scanned archive | Classify first, then standard and OCR routes | Scans do not become empty Markdown; each rerouted file remains linked to its source. |
| Batch containing an invalid file | Continue valid items and isolate failure | Completed outputs remain downloadable; the invalid item has an actionable status and bounded retry behavior. |
This benchmark is an acceptance matrix, not a claim that every PDF reaches the same accuracy. Test representative files from your own collection before automating a large queue.
Batch conversion limits and operational risks
Large queues amplify naming mistakes, duplicated work, and unnoticed extraction errors. Add controls before increasing volume.
- The current client selects at most 20 PDFs, 50 MB per file, and 250 MB total; server entitlements may be stricter.
- OCR-heavy batches consume more processing time and capacity than selectable-text documents.
- Retries should be limited to recoverable failures so invalid files do not loop indefinitely.
- Downloads, assets, retention, and destination folder naming need a deterministic convention before migration.
Frequently asked questions
Batch PDF to Markdown questions
1How many PDFs can I convert in one batch?
The current client allows up to 20 selected PDFs, 50 MB per file, and 250 MB total. Account or backend limits can be different, so the live interface and plan remain the operational source of truth.
2Does one failed PDF stop the entire batch?
The batch model tracks item-level status, so completed files can remain available even when another item fails. Eligible failed jobs can be retried without treating the entire collection as one opaque request.
3Can a batch contain both scans and text PDFs?
It can, but classification improves results. Test normal extraction first, route image-only files to OCR, and apply a stronger review sample to scans, forms, tables, and unusual layouts.
4How should batch output files be named?
Use deterministic names derived from the source, avoid collisions, preserve a manifest, and decide whether assets live beside each Markdown file or in a scoped subfolder. Do not rely on manual download order.