The file your supplier sent is not a catalog. It's a starting point.
Supplier feeds arrive with empty attributes, three vocabularies for the same field, dimensions in mixed units and specs buried in PDFs. Synaps reads those sources, produces the missing values and turns the file into data a channel will accept.
The problem
Every supplier sends a different shape of data
Column names, units, category vocabularies and completeness levels differ per supplier, so a multi-supplier catalog is several incompatible catalogs in one file.
The values that matter are in documents, not fields
Dimensions, materials and technical specifications are frequently present only in a spec sheet or installation PDF, which no import job reads.
Onboarding a supplier is the growth ceiling
Adding assortment is how a catalog grows, and manual review of each new supplier's file is what caps how fast that can happen.
The long tail arrives worst and stays worst
Major brands publish good data. Unbranded, private-label and long-tail products — usually the higher-margin part of the catalog — arrive with the least.
Everything you need
Any format, no custom integration
Spreadsheets, CSV exports, API feeds and flat files are ingested as they arrive, so onboarding a supplier does not require building a connector for them.
Extraction from documents
Specifications are pulled from spec sheets, PDFs and product images, which is where the values missing from the feed usually are.
Vocabulary normalisation
Three suppliers writing colour, hue and shade are reconciled into one attribute, so filters and search behave predictably.
Unit reconciliation
Measurements expressed in different units and notations are resolved to a single normalised value with the unit preserved.
Identifier validation and dedupe
GTINs and MPNs are validated and duplicate variants across suppliers are collapsed, so the same product does not enter the catalog three times.
Completeness scoring per product
Each record is scored against what the destination channel requires, so what is still missing is visible before publication rather than after rejection.
How it works
- 01
Ingest the file as it comes
Synaps takes the supplier's own format without asking them to change it, which is usually the only realistic option.
- 02
Normalise to your standard
Fields, units, identifiers and category values are reconciled to one model, so a multi-supplier catalog behaves like a single one.
- 03
Fill what the supplier left empty
Missing attributes are sourced from documents and product sources, with the origin recorded rather than guessed.
- 04
Score before publishing
Each product is measured against the destination channel's requirements, so incomplete records are held back instead of being rejected downstream.
What to do when supplier product data is incomplete
Incomplete supplier data is product information that arrives missing the attributes a sales channel requires: empty dimensions, no material, absent identifiers, inconsistent categories, specifications present only in an attached document. Making it channel-ready means normalising what is there, sourcing what is not, and validating the result against what the destination channel actually requires. Synaps performs those three steps on the supplier's own file format.
Asking suppliers to send better data is the intuitive fix and rarely the workable one. Suppliers serve many customers, their systems predate your requirements, and the leverage to impose a format usually is not there. Treating incoming data as a raw input to be completed, rather than as a deliverable to be enforced, is what makes multi-supplier assortment scale.
The economics are the reason this changed recently. Reading a supplier document, resolving that two notations describe the same measurement, and deciding what a specific field should contain used to be human work priced per SKU. That work is now largely automatable, which makes it viable to complete the long tail rather than only the products that already came with good data.
One caution worth stating plainly: generating text over an empty record is not enrichment. A generator handed a part number and a two-word title will produce a fluent description containing no facts, and a confident wrong value is worse than an empty field, because an empty field is visibly missing and a fluent sentence is not. Sourcing the value comes first; writing about it comes second.
Frequently asked questions
Send us the worst supplier file you have
Not your cleanest one. We'll return it enriched, scored against a channel of your choice.
Send a file