Welcome to a new edition of Business Case — where we take a fictional but realistic company scenario, run the numbers, and show you what the cost of inaction really looks like.
In manufacturing, the purchase order is where strategy quietly meets reality. Every purchase — a replacement bearing, a contractor, a custom casting — starts as a small request and ends as money leaving the company. Between the two sits the procure-to-pay process: the chain of approvals, contracts, and confirmations that is supposed to make sure the company buys the right thing, at the right price, from the right supplier, fast enough not to stop the line. When that chain works, nobody notices it. When it does not, the cost shows up everywhere except on a report that anyone is actually reading.
This is not a sourcing problem and it is not a price-negotiation problem. It is a process problem. The companies that buy well under pressure do not have tougher buyers or better contracts than everyone else. They have a purchase approval workflow that makes the disciplined decision the fast decision — where routine orders clear in hours, exceptions get the scrutiny they deserve, and every approval leaves a trail that finance can actually read. Today's business case walks through what the absence of that looks like, and what it costs a mid-sized manufacturer that, on paper, is doing just fine.
Meet Reinhardt Antriebstechnik GmbH: a Stuttgart-based manufacturer of industrial drive and gear systems, founded in 1989, today operating three plants across southern Germany and supplying machine builders, automotive tier-one suppliers, and intralogistics OEMs. Reinhardt employs 1,150 people and generates 380 million euros in annual revenue, with an operating profit of around 30 million euros. By every conventional measure the company is solid: a healthy order book, a strong engineering reputation, a broad and stable supplier base.
But the head of procurement and the CFO share a problem they cannot quite put their finger on. Reinhardt spends roughly 220 million euros a year with external suppliers — about 58 percent of revenue — across direct materials and indirect categories like maintenance, tooling, and services. That spend flows through about 95,000 purchase orders a year, raised by buyers across three plants and a central procurement function. The average purchase order takes 11 working days to move from request to approved order; comparable manufacturers with a digitised process do it in three. The fully loaded cost of processing one order sits near 78 euros; the benchmark is closer to 42. Roughly 23 percent of total spend happens off-contract, against a best-practice target below 10 percent. New suppliers take 38 days to onboard, where 12 would be normal. And about 31 percent of orders are touched more than once before they are clean.
None of these numbers is alarming on its own. Together they describe a procure-to-pay process that has grown by accretion rather than design, and that nobody fully owns. The board has asked the same question that started every business case in this series: how much is this actually costing us, and what would it take to stop it? Answering that means treating purchase approval not as administrative plumbing but as the mechanism that decides whether the company's buying power ever reaches its income statement.
When Reinhardt's operations team mapped how a purchase order really moves through the company today, three structural problems came into focus. None of them is unique to Reinhardt. They are the standard failure modes of a manufacturer that has scaled across multiple sites without rebuilding its approval logic.
The first problem is that approvals live in inboxes and signatures, not in a system. A buyer at the Heilbronn plant needs a 40,000-euro tooling order. He emails the plant manager, who replies "go ahead" from his phone. The order is entered into the ERP at the agreed price. Nobody captured why this supplier was chosen over the framework supplier, nobody recorded that the same tooling category is already under contract at a lower rate at another plant, and nobody flagged that the approver was, strictly, two thresholds below his authority for a one-off purchase of that size. Internal review estimates that 64 percent of purchase orders above 5,000 euros have no system-captured approval trail beyond an email thread. For smaller indirect buys, plant staff order against expired contracts or no contract at all, on personal judgement, with no check against negotiated terms.
The second problem is that there are no consistent thresholds, which is the engine of maverick spend. Reinhardt has an approval matrix. It distinguishes order values, categories, and roles, and it looks perfectly reasonable as a document. In practice, each plant interprets it differently. Heilbronn escalates anything above 10,000 euros to the plant manager; Reutlingen escalates only above 25,000; the central team has an unwritten understanding that "urgent" orders skip the matrix entirely so the line does not stop. Buyers quickly learn which approver says yes the fastest and route their harder orders accordingly. The result is that two near-identical purchases receive completely different scrutiny, off-contract buying becomes the path of least resistance, and 23 percent of the company's spend ends up outside the agreements that procurement worked hard to negotiate.
The third problem is that there is no reporting on patterns. Finance sees the aggregate spend number each month. Procurement sees the savings it negotiated on paper. Nobody owns the process-level questions: which approval paths are slowest and lose the most time, which categories leak off-contract, which suppliers are duplicated across plants, which orders bounce back for rework and why. When process mining is finally run on the order data, it shows that the average purchase order is touched by six people, that 19 percent are reworked at least once, and that 220 suppliers deliver a category that 30 could comfortably cover. Until that analysis exists, every one of those orders looks normal on the dashboard.
These three problems compound, and the cost surface they create is larger than it looks. Start with processing. Across 95,000 annual purchase orders, the fully loaded handling cost averages 78 euros against an achievable 42; applying that 36-euro gap conservatively to only the 70 percent of volume that is realistically addressable in a first phase yields roughly 2.4 million euros a year in excess processing and rework. Add maverick spend: 23 percent of 220 million euros is off-contract, and moving even the reducible portion — the gap between today's 23 percent and a 10 percent target, about 28.6 million euros of spend — onto negotiated terms, at a conservative 6 percent price premium, recovers around 1.7 million euros a year. Then add the cost of slowness itself: missed early-payment discounts, expedited freight on rush orders that should never have been rushed, and production disruption while suppliers take 38 days to onboard, together worth an estimated 1.3 million euros annually. The combined cost of weak procurement governance at Reinhardt is on the order of 5.4 million euros a year — roughly 18 percent of the company's operating profit. The board's unease is well founded.
The path forward at Reinhardt is not exotic. It does not require a new procurement team, a different ERP, or a wave of supplier terminations. It requires three integrated levers that turn purchase approval into a structured, visible, fast process. Each lever maps directly onto one of the three problems above. (For a closely related view on how the same dynamics quietly erode margin on the sales side, see our recent business case on how to stop margin erosion in retail sales.)
The first lever is a digital purchase approval workflow with a captured audit trail. Every requisition above a defined threshold is created in one system. The system records the requester, the category, the supplier, the contract reference, the justification the buyer is required to enter, the approval path, and the timestamps at every step. Routine orders that sit inside an existing contract and below the relevant threshold flow through automatically and reach the supplier in hours. Orders that breach a threshold or sit off-contract are routed to the right approver with full context attached, so the approval is a real decision rather than a reflex. The supplier rarely feels the change, because the standard cases now move faster than before; only the exceptions slow down, and they slow down on purpose. For Reinhardt, capturing this trail is expected to cut the undocumented-approval rate from 64 percent to under 15 percent within the first year, and to bring processing cost down toward the benchmark on the addressable volume. The recovered cost from this lever alone is conservatively around 1.8 million euros a year. It costs roughly 150,000 euros in licences and integration in year one, and about 50,000 euros a year to run.
The second lever is a structured approval matrix with catalogue and contract enforcement built into the workflow. The paper matrix is rewritten with clear, narrow thresholds — who can approve what, in which category, at which plant — and then encoded in the workflow engine so that interpretation is no longer a local decision. A 10,000-euro maintenance order against an existing framework clears with one approval; the same value off-contract routes to category management with the contract gap flagged. Catalogues and punch-out connections make the on-contract option the default and the easy one, while off-contract orders require an explicit, recorded justification. Hard rules block the genuinely unauthorised regardless of who tries to wave them through. The effect on behaviour is immediate: buyers stop hunting for the fastest yes, because the system, not the manager, now decides the path. Reinhardt's pilot brought maverick spend from 23 percent toward 11 percent in the categories covered. Annualised across the addressable spend, this lever recovers approximately 1.4 million euros a year, at around 90,000 euros in year one and 40,000 euros a year thereafter.
The third lever is process mining on the procure-to-pay data, feeding a monthly governance review. Once orders flow through one system, the data becomes analysable, and the patterns that no human dashboard would surface become visible: which approval paths take the longest and stall the most orders, which categories leak off-contract, which suppliers are duplicated across plants, which rework loops repeat every month. This is the domain of process intelligence tooling — the category in which noreja operates, and which is catalogued alongside peers on directories such as topai.tools. The output drives a focused, one-hour monthly meeting between procurement and finance, where the worst patterns are addressed by name rather than in aggregate: a supplier base rationalised, a slow approval path redesigned, an early-payment discount finally captured. Within six months, Reinhardt's pilot reclaimed missed early-payment discounts, cut expedited freight, and shortened supplier onboarding materially. The recovered value from this lever is estimated at around 0.9 million euros a year, on an analytics module and governance cadence costing roughly 70,000 euros fully loaded. The principle behind it is simple: you cannot govern a process you cannot see.
Combined, the three levers are expected to recover roughly 4.1 million euros of the 5.4 million euros annual cost in steady state — about three-quarters of the leak. First-year implementation, including licences, integration, training, and process redesign, comes in at approximately 380,000 euros, with annual run costs around 160,000 euros thereafter. Payback is reached inside the first year, typically within the first few months as the rollout completes, and the steady-state contribution to operating profit is in the order of 13 percent above Reinhardt's current baseline. The point is not that Reinhardt becomes a different company. The point is that the buying power the company already has finally shows up where it belongs.
What share of the purchase orders your company approved last quarter would still be visible if you could only see the ones with a system-captured approver, a contract reference, and a written justification?
How much variance is there between your fastest-approving and your most disciplined approver, on objectively similar orders — and if you cannot answer that, what is it costing you not to know?
How much of your spend runs off-contract today, and is that because the off-contract option is genuinely better, or simply because it is the path of least resistance?
If a buyer needs something urgently tomorrow, does your process give them a fast, governed route — or does it force them to choose between speed and discipline?
What would your operating profit look like if even half the gap between your current procure-to-pay process and the benchmark closed, and what could you fund with the difference?
Slow, undocumented procurement is rarely the result of a single bad decision. It is the result of thousands of small orders that nobody had the time, the data, or the authority to route properly in the moment. The fix is not a tougher procurement policy or a louder savings target. The fix is a purchase approval workflow that turns buying discipline from a matter of individual judgement into a property of the system. Once the workflow captures the order, the matrix routes it, and process mining exposes the pattern, procurement stops being something finance reconciles after the fact and becomes something the organisation can actually steer.
We invite you to look at your own numbers with Reinhardt's lens. Measure the average time and cost of a purchase order in your business, and the share of spend that runs off-contract. Trace ten recent orders end-to-end and ask whether the chain of approvals would survive a serious audit. Then decide whether you can afford to leave that gap open for another year.
The procure-to-pay process is the full chain from a purchase request through approval, ordering, and payment. It matters because most procurement value is won or lost not in the contract negotiation but in the day-to-day approvals: a slow or undocumented process lets spend drift off-contract and inflates handling cost long after the savings were negotiated on paper.
A digital purchase approval workflow captures every order in one system with the approver, contract reference, and justification recorded. Routine orders inside an existing contract clear automatically in hours, while exceptions are routed with full context. This cuts processing cost toward benchmark, reduces rework, and gives finance an audit trail it can actually read.
Maverick spend is purchasing that happens outside negotiated contracts, often because the off-contract route is simply faster. It is expensive because those purchases forgo negotiated pricing — typically several percent above contract terms — and because nobody can see the pattern. Moving even the reducible portion on-contract usually recovers more than any single price negotiation.
Process mining analyses the full order data to surface patterns no dashboard catches: which approval paths stall the most orders, which categories leak off-contract, which suppliers are duplicated across sites. The output drives a focused governance review where the worst patterns are fixed by name, turning procurement from something reconciled after the fact into something the organisation can steer.
For a mid-sized manufacturer, a pragmatic implementation typically runs three to five months from kickoff to full rollout, with a pilot covering part of the volume in the first weeks. The investment is dominated by integration to existing ERP systems, and payback inside the first year is normal when the underlying leakage is material.