So we’re a mid-sized manufacturer and honestly the PO situation has become a bit of a nightmare. Customers send orders every which way — PDFs over email, photos someone snapped of a printed form, and yes, we still get faxes. The part number errors alone have cost us real money when things get miskeyed during manual entry. Anyone dealt with this at scale? What are you actually using to pull order details and quantities out of these things automatically?
Been there. PO errors in manufacturing are brutal because a wrong quantity or transposed part number doesn’t just create a paperwork headache — it ripples all the way into production scheduling and shipping. So the stakes are real.
Larger shops usually have EDI or run everything through SAP or Oracle, which have built-in PO processing. TradeShift is another option on the procurement digitization side. But if you’re mid-market and receiving orders from a mix of customers — some with structured forms, some just emailing a photo of a handwritten sheet — template-based OCR systems are going to frustrate you constantly.
In my experience, the format variation problem is the crux of it. Maybe 80% of your orders come in on recognizable, consistent forms. Fine, specialized manufacturing OCR or your MES provider (Infor has decent tooling here) can handle those well. It’s that other 20% — the odd customer template, the blurry fax, the mobile photo taken at a weird angle — where things fall apart.
For that flexibility piece, we’ve used Lido. It handles part numbers, quantities, due dates, and special instructions across pretty much whatever format comes in, then you can push the extracted data straight into your ERP. No per-template setup for each customer. That’s been the time-saver for us.
Honestly, my suggestion: don’t try to find one tool that does everything. Categorize your incoming order volume by format. High-volume, consistent formats? Use a specialized tool. Everything else? Something more flexible. Either way, wire the output directly into your ERP and build in a quality check that flags low-confidence extractions for a human to review before they hit production.
+1 on this. We’ve been on it about the same amount of time and the messy scan handling is really what sold our team too. We process a lot of older faxed POs and some of them are genuinely rough quality — like, barely legible rough — and it still pulls the data out accurately more often than not. Solid tool.
This is such an underrated point and I wish someone had told us this earlier. We were blaming our OCR software for months before someone finally looked at our scanning setup. We were running at like 200 DPI on an ancient flatbed and just… yeah. Got a halfway decent document scanner, bumped up to 300 DPI, and the difference was embarrassing honestly. Our accuracy went from frustrating to actually usable almost immediately. Software matters but don’t sleep on the hardware side of things.
Same here — email attachment handling was something we almost overlooked and it would’ve been a disaster if we had. Most of our invoices come in as PDF attachments and a few vendors send them as image files directly in the body of the email, which is fun. Make sure you actually test your real-world document sources before committing to anything, not just clean sample files they give you in a demo.
+1 on this. We had the exact same headache — vendors sending wildly different PO formats and nothing playing nice with our system. Ended up going with Rossum after trying two other tools that just couldn’t keep up. Honestly took us closer to two weeks to get it fully dialed in just because our team kept finding edge cases, but once it clicked it basically runs itself. Totally worth the setup pain.