Supplier PDF Data Sheets to PIM: What Your Import Can Do

Disclosure. openProd publishes this page, and openProd is one of the onboarding tools it mentions. Each PIM is described only with what its vendor states in its own documentation, with a link to the source. Checked on 13 October 2026. If something has changed, tell us and we will update the page.

The short answer. Most PIM imports read structured files, not PDFs. Pimcore’s Data Importer, Akeneo’s import profiles and Plytix import feeds take CSV, XLSX, XML or JSON. Ergonode’s AI Complete can use an uploaded PDF to fill chosen fields, and Inriver added PDF onboarding in spring 2026. Everywhere else, someone or something has to turn the data sheet into values that fit your model, and you should be able to check where each value came from before it is saved.

Why supplier PDFs are hard to import

A data sheet is made for a person to read, not for a system to import. The values are there, but their meaning often sits in the layout around them. The same patterns come up in almost every batch:

• Tables split across pages. A specification table starts on page 2 and continues on page 3, sometimes with the header repeated, sometimes without it.
• Units in the header, not in the cell. A column called “Weight (kg)” holds 0.09. Copy the cell alone and the unit is gone.
• Footnotes and conditions. A value marked with an asterisk may hold for one voltage or one finish, and the condition sits at the bottom of the page.
• Several products on one sheet. One data sheet often covers a whole range: one column per model, or one row per size. Each value has to reach the right product.
• Supplier wording. The supplier writes “Housing material: Stainless steel”; your dictionary expects a code such as stainless_steel. The supplier writes “IP rating”; your attribute is called “Protection class”.
• Scanned pages. Some PDFs are pictures of paper. There is no text to copy, so the text has to be recognised first, and a misread digit is easy to miss.
• Other languages. A data sheet in German or Italian still has to land on your attributes and option lists.

None of this is unusual. It is ordinary work in a catalogue team, and it is why “import the PDF” is rarely one click.

What your PIM’s import takes

Before you pick a workaround, check what your PIM already does. Below is what each vendor states in its own documentation, checked on 13 October 2026. “Not stated” means we found no public statement, not that the feature is missing.

Pimcore. The Data Importer lists six data sources (asset, upload, HTTP, SFTP, push endpoint and SQL) and five formats (CSV, JSON, XML, XLSX and SQL query results), with a mapping editor that previews the source data and every transformation step. PDF is not mentioned. On Pimcore, a data sheet has to become one of those formats first, or come in through an extension.

Akeneo. Akeneo PIM imports run through import profiles and accept CSV and XLSX; Tailored Import maps XLSX or CSV files whose structure differs from what the PIM usually accepts. Supplier Data Manager imports structured files (xls, csv, xml and others) with their assets. Its extraction module fills missing attributes from configured text fields and from image assets of type media, and flags low-confidence values for manual review. PDF as an extraction source is not stated in that documentation.

Ergonode. Ergonode imports XLS files through a mapper UI and can update existing products, create new ones, or both. Ergonode Transfer exchanges data as CSV files packed in a ZIP archive. PDFs come in through AI Complete: you select which fields to complete and upload a PDF file, or choose reference attributes, to guide the AI.

Plytix. Automated import feeds accept CSV from a URL, FTP or SFTP, or Dropbox. AI Autofill works on short text, paragraph and HTML attributes from a prompt and referenced attribute values, and you apply or discard the result. Neither help article mentions PDF.

Inriver. The Spring 2026 release added unstructured data onboarding for PDF, Excel, Word and PowerPoint, with AI extraction and field mapping and a final human review step before data is committed. The data goes into Inriver.

Structured imports are mature in every one of these products. PDF as a data source exists in some of them, in different shapes: filling selected fields on a product (Ergonode) or a document intake into the PIM (Inriver). If you depend on it, check the current documentation for your edition.

Four ways teams get PDF data in today

Retyping

Someone opens the PDF on one screen and the PIM on the other, and types. It needs no tools and works for a few products a month. It is also where unit slips and swapped digits happen, and once the value is in the PIM, nobody can tell which page it came from.

What to check: a second person checks the numbers, and the product record notes which file and version it came from.

Copying into a spreadsheet, then importing

Tables are copied from the PDF into a spreadsheet, cleaned up and loaded with the PIM’s own import. This uses tools you already have, and the PIM import does the final mapping. The cleanup is the real work: merging tables split across pages, moving units out of headers, and turning supplier wording into your option codes.

What to check: a column for the source file and page, units converted before import rather than after, and a test import on a few products before the full batch.

Generic text recognition or a general AI assistant

Text recognition turns a scanned page into text. A general AI chat assistant can go further and return a table from a PDF. Both are quick to try. Neither knows your data model: the output uses the supplier’s names, units and wording, not your attributes and codes. The link from a value back to its place in the PDF is lost, and there is no review step or record of who approved what.

What to check: every number against the PDF, especially decimals and units; whether the output invented a value the sheet does not contain; and where the file is sent and stored.

An onboarding tool in front of the PIM

A dedicated onboarding tool reads the PDF, maps values to your attributes, units and option codes, asks your team about what is uncertain and hands checked data to the PIM. It is built for the gap between the data sheet and the import. It is also one more tool to evaluate. Our comparison of product data onboarding tools lists what each vendor states about PDFs, review and source per value.

What to check: the points in the next section, on your own files.

What to look for in a tool that reads PDFs

Whatever you choose, these six things decide whether you can trust the result:

1. A link from every value to its source. You should be able to open the PDF at the place a value was read, not just see the file name.
2. Mapping to your model. Values should land on your attributes, units and option codes, not on a generic template you then map again.
3. Unit conversion with the value as read visible. If 0.09 kg becomes 90 g, you should see both before saving, so a wrong conversion is caught by a person.
4. Values without a place, listed. A data sheet often holds features your model has no attribute for. They should be listed, not dropped silently and not forced into the nearest field.
5. A preview before saving. New products and updates should be separated, with each changed value shown as it is now and as it will be after import.
6. Change history. Every saved change should be recorded, with a way to see what changed and restore an earlier version.

How openProd reads supplier PDFs

openProd is an AI onboarding layer in front of your PIM. It reads supplier PDFs (product names and specification tables), spreadsheets such as price lists and update files, and text pasted from an email. Files can also come from the openProd asset library.

Before the AI analysis starts, openProd shows a credit estimate based on similar imports, with the expected time, and the analysis runs after you accept it. The analysis proposes data; your catalogue is not touched.

openProd suggests the product family and shows how many attributes match; nothing is set until you choose. Each value is then matched to an attribute of your own catalogue and lands on your units and dictionary codes. A weight read as 0.09 kg is saved as 90 when the attribute expects grams, and “Stainless steel” becomes stainless_steel, the code your dictionary uses.

Clear values go straight through. Conversions and values without an attribute wait for a decision in Data review (Decide). Show in source opens the PDF at the place the value was read. You can compare the value as read with the value after conversion, save or skip it, and step back one decision. Values your model has no attribute for are listed separately, and deciding on them is optional.

The preview separates new products from updates and shows each changed value, now and after import. Nothing is written until someone clicks Save to catalog, and after saving openProd confirms that the saved data matches the checked result. Every product change is recorded, so you can see what changed and restore an earlier version.

From there, the Pimcore connector and the Ergonode connector hand the checked data to your PIM; for Akeneo and other systems, the connection is agreed with your team and tested in a pilot. See AI import for the steps, and our guides for Pimcore and Ergonode teams.

Which route fits your setup

• A few PDFs a month from a stable supplier: retyping with a second check may be enough.
• Suppliers can export structured data: ask for a spreadsheet or XML and use your PIM’s import (Pimcore Data Importer, Akeneo Tailored Import, Ergonode XLS import).
• You run Ergonode and need a few missing fields per product: try AI Complete with the uploaded PDF.
• You run Inriver: test unstructured data onboarding on your own data sheets.
• You run Akeneo and suppliers will use a portal: see our comparison of Akeneo Supplier Data Manager and openProd.
• Many suppliers send their own PDFs, and you keep your PIM: compare onboarding tools in front of the PIM on your own files.

A checklist for your next PDF batch

1. Pick three real data sheets: one simple, one with a table across pages, one covering several products.
2. Write down, for a few products, the values a careful person would enter, in your units and codes.
3. Run the sheets through your current route or the tool you are testing.
4. Compare every value with your list. Count the wrong values that nobody flagged, not just the ones that were flagged.
5. Check that units, option codes and decimals match your attributes.
6. Open the source of a few values and confirm the page and table are right.
7. Look at what happened to values your model has no place for.
8. Check what will be written to the PIM, and when, before you approve it.
9. Check that the change is recorded and can be restored.

If you want to see your own data sheet go through this, book a demo and bring one of your supplier files.

Sources and updates

• Pimcore Data Importer: docs.pimcore.com
• Akeneo PIM imports: Import your data, Tailored Import
• Akeneo Supplier Data Manager: product page, extraction module
• Ergonode: features, Ergonode Transfer files format
• Plytix: import feeds, AI Autofill
• Inriver: Spring 2026 release notes, release announcement

Checked on 13 October 2026. If something has changed, tell us and we will update this page.