Supplier data & capture7 min read
Where product data actually lives
Before you can automate product data capture, you need to be honest about where it comes from today. There are usually seven sources, and they are not equally difficult.
ReadBefore you can improve how product data reaches your catalogue, you need an honest inventory of where it comes from today. In most distribution businesses there are seven sources: the supplier's public website, PDF catalogues, spreadsheets from the supplier, dealer portals, email attachments, manufacturer syndication feeds, and the institutional knowledge held by staff.
They are not equally difficult, and the ranking is counter-intuitive. The supplier website feels like the most manual source and is in fact the most automatable, because it is internally consistent — work out how one manufacturer structures their product pages and that knowledge applies to every product they sell, until they redesign. The PDF catalogue feels structured and is not: a PDF describes where ink goes on a page, so the table you can see is not a table to a machine.
A source automates well when it is consistent, addressable and complete. Supplier sites usually satisfy all three. PDFs satisfy the first two and fail the third in a specific and dangerous way — the information is present visually but not structurally, and when extraction breaks, nothing in the output tells you it broke.
Two sources will never automate and you should stop trying. Email attachments need a process, not an extractor. Institutional knowledge has to be interviewed out of people and written down as compatibility relationships and cross-references — a genuine project, and the one that most separates a good catalogue from a complete one.
Guides in this topic
Supplier data & capture8 min read
Why supplier PDFs are the worst source of product data
A PDF is a picture of a table. Everything downstream of that fact is harder than it looks.
ReadCommissioned and scheduled
See it against your own catalogue.
Thirty minutes. Bring three supplier URLs and we will capture them live, so you are judging the output rather than the pitch.