Product data entry, without the double-keying.

Product data entry is the dullest job in ecommerce and one of the costliest when it goes wrong. A wrong size on a variant becomes a return. A missing material field becomes a rejected feed item. A price typed with a misplaced decimal becomes a very bad afternoon.

Scenox Commerce handles product data entry as a managed service. You send source material, and you receive structured, store-ready data.

What you get

• Source data extracted into your platform's exact import format, whether Shopify, WooCommerce, Magento, BigCommerce, Wix or a custom build
• Titles, SKUs, prices, weights, dimensions, barcodes and stock levels mapped to the right columns
• Variants and options built from the supplier's size and colour logic
• Attribute values standardised: units, capitalisation, colour names, materials
• Images matched to the right product and variant
• A QA report of fills, flags and unresolved items

How it works

1. You send spreadsheets, PDFs, images or a store export, and tell us the target platform.
2. We inspect the files and send a fixed quote and a written delivery date.
3. The AI engine extracts and structures the data; a person reviews every output against the source documents.
4. You approve in a draft or staging state, with a list of any flagged items for your decision.
5. We load the data through a staff account, or hand over a ready-to-import file.

Why AI plus review beats manual double-keying

The old defence against typing errors was double-keying: two people enter the same data and a system flags the differences. It works in principle. In practice it doubles the labour, it is slow, and both operators can make the same mistake when the source is ambiguous, such as a smudged PDF or a supplier abbreviation.

Our process changes where the human effort goes.

Extraction is not typing. The AI engine reads the source and produces structured fields directly. Nobody retypes a 14-character SKU, so nobody transposes two characters in it.

Rules run on every row. Before a person sees anything, the data is checked against rules: required fields present, prices numeric, variants sharing consistent option sets, units standardised, duplicates flagged.

The reviewer checks, rather than creates. A person compares output against the source and focuses on the cases the engine marked as uncertain. Reviewing is faster than keying and catches a different class of error: values that are formatted correctly but wrong.

Ambiguity goes to you. If a PDF says "approx. 30cm" or two suppliers give different weights, we do not guess silently. It appears in the QA report as a question for you.

We do not publish accuracy percentages, because they depend entirely on how clean your sources are. The QA report gives you the real picture for your own data, product by product.

Inputs we deal with every week

Accuracy checks and what we will not guess

Every field has a source: your supplier sheet, spec PDF, packaging photo or an existing store record. Where a value is missing, we leave it empty and list it in the QA report rather than invent it. Checks we run on each batch include required fields filled, units and formats consistent, prices numeric and in the right currency, SKUs unique, barcodes valid length, image links resolving and variants attached to the correct parent.

Typical batch deliverables are the filled import file in your platform's format, an exceptions list, and a short note on anything ambiguous, such as two supplier rows that look like the same product. You decide those; we do not merge silently.

We keep the work to product data, and we never use your files to train third-party models.

Pricing

Starting prices for catalog listing, which includes data entry and upload:

The fixed quote comes after we inspect your files, before work starts. See pricing, or read what drives the numbers in how much product upload costs. The service is described in full at bulk product upload.

Is this a fit?

It suits you when you have hundreds or thousands of products to enter, source files in awkward formats and no spare staff hours. Retailers adding a new supplier line, brands relaunching on a new platform and agencies with a client backlog all fit well.

If you have thirty products and a tidy spreadsheet, enter them yourself; the time saved will not justify a project. And if your source data is incomplete, no process can invent what is missing. We will tell you which fields are absent before we quote, and you decide whether to supply them or leave them blank.

Try the free 10-product sample if the project is a fit, or ask for a fixed quote at /get-started.