How to automate supplier price list imports into PIM without breaking the catalog

Learn to parse supplier Excel and CSV files, map fields, validate categories, and safely run PIM imports with a dry run.

  • Why a single template does not solve the problem
  • Import draft before writing to the catalog
  • How controlled import works
  • Where the agent’s authority ends

Why a single template does not solve the problem

Suppliers send price lists in Excel, CSV, and other formats. In one file, the SKU is called “Code”; in another, “Supplier SKU.” Headers may start on the second row, categories may span several columns, and photo links may be scattered throughout the file.

Even one supplier may change its structure after updating its accounting system. An integration tied rigidly to column numbers therefore breaks at the first change. AI is useful as an interpretation layer: it prepares a verifiable configuration, while PIM and an employee control data writes.

Import draft before writing to the catalog

AI price list import agent

Connect a new price list: categories in three columns, images in two fields, and headers starting on the second row.

Import settings

Interface concept
Structure
headers - row 2, data - from row 3
Categories
three-level hierarchy
Required fields
SKU, name, price
Media
primary and additional image

Agent checked

The agent identified the file structure and mapped columns only to permitted PIM fields. Existing supplier settings were preserved.

  • Source columnsall suggested columns exist
  • PIM schematarget fields are permitted
  • Pricecalculation formula was not changed by the model

Suggested action

Run a test import and show problematic rows

After confirmation: PIM: supplier profile · awaiting review

Trace: price list + catalog schema + supplier profile + execution log

Fix settingsRun dry-run

How controlled import works

  1. 01

    Determine structure

    The agent identifies the sheet, encoding, delimiter, header row, and start of the data from limited samples.

  2. 02

    Map fields

    Columns are assigned only to permitted fields: SKU, name, price, images, and attributes.

  3. 03

    Restore categories

    The system supports a single category, categories from columns, and a mixed hierarchical catalog.

  4. 04

    Review dry-run

    Before loading, you can see missing values, errors, unknown categories, and planned creations or updates.

  5. 05

    Confirm and save

    An employee starts the import, while the settings and result are saved in the profile and history.

Where the agent’s authority ends

SituationWhat the agent doesWhat the employee or system does
Unknown structureSuggests a parsing methodChecks the result
New columnMaps to a permitted fieldConfirms the assignment
Unknown categorySuggests a catalog branchDetermines whether to create a category
Price calculationIdentifies the source columnApplies the approved formula
Actual importPrepares a validated draftStarts loading

Assess where AI can deliver impact in your process

Safe import workflow

A heterogeneous file becomes a verifiable PIM change

Input

Excel, CSV, linkfile and supplier profile

AI analysis

Structure and mappingsonly permitted fields

Control

Dry-runerrors, categories, and future changes

Solution

Employeecorrects or confirms

Recording

PIM and logproducts, profile, execution history
AI handles file ambiguity, while programmatic checks and a human protect the catalog from uncontrolled changes.

Why prices should not be delegated to a language model

The model may determine that the “MSRP” column corresponds to the retail price. But the final price depends on the currency, VAT, markup, region, customer segment, and offer validity period.

Therefore, AI identifies the purpose of the source field, while calculations follow formal rules. This separation makes every final price explainable and repeatable, regardless of the prompt wording.

What to measure during the pilot

T₀→T₁time from receiving the price list to a ready-to-run import
M₀→M₁number of manual mappings
E₀→E₁errors and reimports
S₀→S₁share of stable supplier profiles

Supplier connections

Design price list imports into PIM

1 flow

We will review real files, define the data model, set up a verifiable draft and dry-run, then determine the conditions for automatic execution.

  • supplier profiles
  • category and field validation
  • pilot metrics
Discuss PIM import

FAQ

Frequently asked questions about importing price lists into PIM

Can price lists be uploaded without employee involvement?

Yes, for stable suppliers with a saved profile. Loading stops if the structure changes, new categories appear, or mandatory rules are violated.

Do we need to send the entire price list to the model?

No. Headers and a limited sample of rows are usually enough to determine the structure; sensitive values can be masked.

What should I do after changing column names?

The agent suggests a new mapping, the system compares it with the profile, and an employee confirms any significant change.

Can the agent calculate prices?

It is better to separate the tasks: the agent identifies the source field, while deterministic business logic applies the formula.

Discuss the article: How to automate price list imports…

Enter your email or phone number so we can get back to you.

Send via: