If you work in procurement at a manufacturer, you've lived this moment. A 2025 survey of 656 manufacturing executives found that only 43% use digital collaboration platforms for supplier data. The rest are still on email and spreadsheets. So someone recommends a procurement platform. You sit through the demo. It looks slick. Catalogs, approval workflows, spend dashboards. Then you ask the one question that kills the conversation:

"Can it handle a multi-part RFQ with engineering drawings, custom specs, and quotes from five different suppliers in five different formats?"

Silence. Or some version of "not exactly."

Most procurement software was built for indirect spend. Office supplies, MRO, IT equipment, professional services. Catalog items with standard pricing where the hardest part is getting the right approval before someone clicks "buy." That's a real problem worth solving. It's just not your problem.

You already know the difference between sourcing raw steel and ordering toner cartridges. You don't need a primer on direct vs. indirect. What's worth talking about is why the software gap is so persistent, and why it's only now starting to close.

The Gap Isn't Complexity. It's Architecture.

The big names (Coupa, SAP Ariba, Jaggaer) built their platforms around indirect spend. That's where the market was when they started, and it's where most of their customers still live.

Over time, they've tacked on "direct materials" modules. But these are afterthoughts grafted onto architecture designed for catalog purchasing and invoice automation. They check the box on a feature list. They fall apart in practice.

Start with the data problem. These platforms expect clean, structured inputs: part numbers in a database, pricing in standard fields, suppliers in a qualified list. Your reality is PDFs with non-standard layouts, specs buried in engineering drawings, supplier qualification data scattered across quality audits and certifications and tribal knowledge in someone's head. The platform wants neat rows and columns. You have a pile of attachments in your inbox.

Then there's adoption. Enterprise platforms need your suppliers to log into a portal. That works for large catalog suppliers who have teams dedicated to maintaining their presence on procurement platforms. Your direct materials suppliers? Often 50-person machine shops without IT departments. They're emailing you a PDF. That's the deal. Ardent Partners' AP Metrics That Matter report shows that manual processing costs $12.88 per invoice versus $2.78 automated, but the savings only materialize if people actually use the system. And they won't.

These platforms also treat procurement as a transaction. Select, approve, order, receive, pay. Direct materials sourcing has back-and-forth on tolerances, negotiations on tooling costs, conversations about alternative materials, haggling over delivery schedules. None of that fits in a dropdown menu. When the tool forces your buyer into a linear workflow for something that's inherently nonlinear, they just go back to email.

And the price tag. Full implementations run six to seven figures, plus ongoing license costs, plus internal IT resources to maintain everything. For a manufacturer with a 10-person procurement team, you'd spend more on the software than you'd save.

The Spreadsheet Is the Real System of Record

Without a Fortune 500 budget, you're stuck. And you know it.

Basic procurement tools can't handle direct materials sourcing. You need RFQ management, quote comparison across unstructured formats, supplier qualification tracking, integration with engineering. The enterprise suites can do some of that, but they're built for companies with 50-person procurement teams, dedicated IT staff, and seven-figure software budgets. Even large companies question whether the ROI is there when adoption stays low.

So most manufacturers use the same thing they've used for 20 years: email, Excel, and whatever ERP they've got. Nobody wants to admit the spreadsheet is the system of record, but it is.

This isn't because procurement teams are unsophisticated. Nobody built a tool that actually fits. The market offered two choices (too simple or too expensive) and teams duct-taped their own workflows out of whatever was available. It works until volume picks up or someone leaves and takes half the institutional knowledge with them.

What Kills a Tool in Three Months

Skip the feature comparison sheet. Here's the stuff that determines whether your team actually uses a procurement tool or quietly goes back to email.

It has to handle whatever format your suppliers send. Your suppliers aren't going to change how they quote. Not for you, not for anyone. The software needs to deal with PDFs, Excel files, email text, scanned documents, and pull out part numbers, unit prices, lead times, MOQs, tooling costs, and terms automatically. If it requires suppliers to fill out a web form, it's dead on arrival. We've talked to dozens of procurement teams about this. It's the single most common dealbreaker.

Reading quotes is table stakes though. A drawing revision drops from engineering while suppliers are already quoting on the old version? That's Tuesday. BOMs, material specs, revision histories, all of that has to flow through. And here's one most tools get completely wrong: BOM-level cost analysis. Supplier A is cheaper on parts 1 through 5, Supplier B wins on 6 through 10. Splitting the award means two sets of tooling costs, two quality qualification processes. The cheapest per-part price might not be the best total cost. You need to see that tradeoff before you make the award, not after.

If you're in aerospace, defense, medical devices, or pharma, supplier qualification tracking can't be an afterthought either. Certifications, audit results, approved process lists, first article inspections. You need to know a supplier isn't qualified before you send them an RFQ, not after they've won the business.

Then there's the channel problem. Every supplier portal that's tried to pull communication onto a different channel has failed. Your suppliers email you. Your team emails suppliers. Work with that or lose. And direct materials pricing isn't a single unit price. You're negotiating quantity breaks at three volume levels, tooling amortization over two years, blanket order pricing with quarterly adjustments, consignment arrangements. If the tool can only capture a single line-item price, it wasn't built for your world.

So Why Is Anything Changing Now?

Honestly? For a long time, nothing was. You can't force direct materials procurement into a transaction-oriented platform designed for catalog purchasing. People tried. It didn't work.

AI is what cracked it open. Reading a messy PDF quote, figuring out which numbers are unit prices versus tooling costs versus shipping, normalizing all of it for comparison. Five years ago that wasn't realistic. Now it is. McKinsey estimates AI can deliver efficiency gains of 20 to 30 percent or more in procurement operations, and direct materials is where the biggest gains are sitting. Gartner predicts that 40% of enterprise apps will have AI agents by end of 2026, up from less than 5% in 2025.

The teams we've talked to that have made the switch aren't going back. Sourcing cycles that used to take weeks are down to days. Buyers spend time on supplier strategy instead of copying numbers between spreadsheet tabs.

Lumari lives inside your email, reads whatever your suppliers send (PDFs, spreadsheets, free-form quotes), and turns it into a normalized comparison without forcing anyone onto a portal. No six-figure implementation to find out if it works.

Sources

1. BusinessWire, "Half of Companies Still Use Email or In-Person Meetings to Share Critical Supplier Data" - https://www.businesswire.com/news/home/20250909900208/en/Half-of-Companies-Still-Use-Email-or-In-Person-Meetings-to-Share-Critical-Supplier-Data

2. McKinsey & Company, "Redefining procurement performance in the era of agentic AI" - https://www.mckinsey.com/capabilities/operations/our-insights/redefining-procurement-performance-in-the-era-of-agentic-ai

3. Gartner, "40 Percent of Enterprise Apps Will Feature Task-Specific AI Agents by 2026" - https://www.gartner.com/en/newsroom/press-releases/2025-08-26-gartner-predicts-40-percent-of-enterprise-apps-will-feature-task-specific-ai-agents-by-2026-up-from-less-than-5-percent-in-2025