Latest / The Buyer's Side with Fexingo: Procurement, Sourcing, and B2B Purchasing Decisions / The Silent Tax of Supplier Invoice Errors
Transcript
- Lucas: There is a number that haunts every chief procurement officer I talk to, and it is not the price of steel or the cost of container shipping. It is four percent. That is the estimated rate at which B2B invoices contain material errors—typos in quantities, mismatched unit prices, or duplicate line items—and yet still get paid without question. Luna: Four percent sounds small until you realize that for a mid-cap industrial company, that might be ten million dollars disappearing into vendor pockets simply because nobody had time to check the math. Lucas: Exactly. And the terrifying part is that we have all this technology now. We have ERPs, we have automated reconciliation tools, we have ai driven matching engines. Yet, according to recent industry surveys from organizations like APQC, nearly half of all invoices still require some form of manual intervention before payment. The friction isn't technical; it's structural. Luna: So you are saying the problem isn't that our systems are dumb, but that our processes are designed to be slow? Lucas: Precisely. Let me give you a concrete example from a major consumer goods manufacturer we will call Apex Consumer Brands. They spent eighteen months implementing a new supplier portal intended to eliminate paper invoices entirely. The goal was straight digital capture and automatic three-way matching against purchase orders and delivery receipts. Within six months, their invoice processing time dropped by forty percent. But their error rate didn't drop. In fact, it ticked up slightly. Luna: That makes no sense. If the data is digital, the system should catch mismatches instantly. Why would errors increase? Lucas: Because the suppliers changed their behavior. When you force suppliers onto a rigid digital platform, many of them don't have the internal tech stack to match your data formats perfectly. So they start entering rough estimates in the free-text fields just to get paid faster. The system sees a free-text description that vaguely matches the PO, auto-approves it, and passes over actual discrepancies in the line-item pricing. Luna: So the automation is creating a false sense of security while letting bad data slip through the cracks. Lucas: It creates what we call compliance theater. Everyone looks efficient on the dashboard, but the underlying financial integrity is rotting. I want to drill into why this happens specifically with mid-tier suppliers. These are not giants like Amazon or Walmart. These are specialized component makers, logistics firms, or raw material providers who are often using legacy accounting software. Luna: Right, and if they are struggling to keep their own books straight, asking them to perfectly format an XML file for your ERP is going to cause more harm than good. Lucas: It causes exactly that. Here is the thing about human psychology in procurement. When an invoice comes in and it is ninety-eight percent correct, the buyer's brain does a quick heuristic scan. It sees the right vendor name, the right total amount, and maybe one wrong quantity on a minor accessory item. The cognitive load to investigate that single discrepancy feels higher than just signing off. Especially if the payment is due tomorrow and there is a late fee penalty involved. Luna: So the incentive structure actually punishes thoroughness. You save time by ignoring the error, and you risk a late fee by fighting it. Lucas: That is the core failure. We have optimized for speed of payment, not accuracy of spend. Let me bring in a specific case study from a global logistics provider, let us call them TransGlobal Logistics. They used to pay twenty dollars per invoice in processing costs. After automating, they thought they were down to five dollars. But then they audited the actual cost of rework. Every time they caught an error downstream in finance—which was often weeks later—they had to pull the supplier back on the phone, issue a credit memo, and re-process the payment. That rework cost thirty-five dollars. Luna: Wait, so the error is cheaper to fix at the point of entry, but by the time it gets to finance, it is seven times more expensive to resolve? Lucas: Seven times is conservative. Some studies put the multiplier at ten to twelve times. The reason is simple: context loss. By the time the invoice hits the general ledger, the original conversation about the scope of work has been forgotten. The buyer who negotiated the deal has moved on. The person reviewing the payment is just looking at numbers on a screen with no memory of the contract terms. Luna: This feels like a classic information asymmetry problem. The supplier knows the truth about what was delivered, and the buyer only sees a sanitized digital record. Lucas: It is, but it is also a data silo problem. Most companies have their purchase order system in one cloud instance, their inventory management in another, and their accounts payable in a third. When an invoice arrives, the system tries to stitch these together. If the SKU codes don't match exactly between the warehouse receipt and the invoice, the system flags it as a mismatch. But often, the SKU code is wrong because the supplier uses a different nomenclature for the same bolt or bearing. Luna: So you end up rejecting valid invoices because of formatting differences, or you accept invalid ones because the system gives up trying to reconcile them. Lucas: You do both. And this leads to what I call the 'lazy approval' phenomenon. Procurement teams start setting tolerance thresholds. Say anything within two percent of the expected cost, and auto-approve it. This saves time, but it systematically transfers wealth from the buyer to the seller. Over a year, those two percent variances add up to millions in overpayments that no one ever questions. Luna: Two percent sounds trivial, but if you are buying ten billion dollars in goods annually, that is two hundred million dollars left on the table. Lucas: Two hundred million is a massive number. And the worst part is that it is invisible. It doesn't show up as a red flag in the audit report. It shows up as slightly lower margins than expected. Executives assume it is market pressure or raw material inflation, when really it is just sloppy invoicing practices being absorbed by the organization. Luna: So how do we break this cycle? Because telling suppliers to be perfect is not realistic, and forcing buyers to check every line item is unsustainable. Lucas: The solution is not better checking; it is better pre-agreement. We need to move the validation upstream. Instead of waiting for the invoice, we validate the expectation at the point of order. I worked with a manufacturing client who implemented a simple rule: any change to the purchase order after confirmation must trigger a revised quote from the supplier before the shipment leaves their dock. Luna: So you prevent the discrepancy from even existing, rather than trying to catch it later. Lucas: Exactly. It shifts the burden of proof to the moment of change. If the supplier changes the material grade, they have to send a new digital confirmation. If they don't, the original PO stands, and any invoice deviation is automatically rejected. It removes the ambiguity. It forces the supplier to align their billing with the agreed scope before any physical exchange happens. Luna: That requires a lot of discipline from the buyer side too. You can't just let shipments happen without verifying the paperwork first. Lucas: It does. But it is cheaper to enforce discipline at the order stage than at the payment stage. Another tactic is standardizing the invoice format at the category level. For commodities like steel or chemicals, you can mandate a specific data schema. For services, like consulting or maintenance, you require a milestone-based breakdown. One size does not fit all, but rigid uniformity across similar categories reduces the friction significantly. Luna: I like that. Treat the invoice not as a request for payment, but as a final verification step in a data pipeline. Lucas: That is the mindset shift. An invoice is a data event. It should carry structured data that matches the purchase order and the receiving report. If it doesn't, the system should reject it immediately, not send it to a human inbox. Humans should only handle exceptions, not routine transactions. Luna: But what about the suppliers who genuinely struggle with technology? Won't this alienate smaller vendors? Lucas: That is a fair concern. If you are too aggressive, you might push them toward competitors who are less rigorous. But here is the counterpoint: most large buyers offer simplified portals or even email to invoice tools that parse basic text. The key is communication. You tell suppliers upfront: 'We pay fast, but we require accurate data.' Speed of payment becomes the reward for accuracy. Luna: So you use payment terms as leverage for data quality, not just price. Lucas: Absolutely. Net thirty days for perfect invoices. Net sixty for ones that require manual correction. It sounds harsh, but it aligns incentives. Suppliers will quickly learn to get it right because cash flow is king. And you reduce your own operational drag. Luna: It is interesting how much of procurement is actually behavioral economics disguised as accounting. Lucas: It is. We forget that behind every invoice is a person trying to get paid, and behind every rejection is a person trying to protect their budget. When we design systems that make honesty easy and lying hard, everyone wins. Luna: Before we wrap up this deep dive into the mechanics of spending, I want to mention something important about how this podcast exists. Lucas: If these conversations have moved your work forward in some small way, consider supporting the network directly. A couple of dollars a month is genuinely what keeps these going — buy me a coffee dot com slash fexingo. It helps us keep the research deep and the ads out. Luna: Yeah, it makes a real difference in keeping the signal clear. Thanks for listening, whether you are a regular or just tuning in today. Lucas: Back to the topic. The final piece of the puzzle is continuous feedback loops. You cannot just set rules and walk away. You need to review the error rates monthly with your top fifty suppliers. Share the data with them. Show them where they are missing. Make them partners in accuracy. Luna: Collaborative correction rather than punitive rejection. That changes the relationship from adversarial to cooperative. Lucas: It does. And when you do that, you stop seeing invoices as noise and start seeing them as a mirror of your supply chain health. High error rates usually mean high complexity, poor communication, or weak relationships. Fixing the invoice fixes the connection. Luna: So the next time an invoice comes in with a typo, don't just pay it. Ask why it happened. Lucas: Ask why it happened, and then ask how to make sure it never happens again. That is how you turn a mundane administrative task into a strategic advantage.