Purchase orders over LINE? Where purchasing breaks between request and receiving

Purchasing has six stages: purchase request, approval, ordering, delivery tracking, receiving, and reconciliation and payment. Small and mid-sized companies rarely have all six broken. Usually it breaks in the same two places: approvals leave no trail, and receiving and reconciliation are recorded separately. This post lays out all six stages and covers what each one actually hands off, which breaks cost you money, and which you can leave alone for now. It also explains why "systemize everything" is the approach most likely to fail, and when you should not touch your purchasing process at all.

Johnny Yang12 min read
Purchase orders over LINE? Where purchasing breaks between request and receiving

"Hold on, let me find that message."

That day I was at a company that makes electromechanical parts. I asked the purchasing manager to pull up one of the larger orders from last month. He opened his computer first. The system had only a bare transaction line. He went through the stack of signed paper on his desk. Not there either. Finally he picked up his phone, opened LINE, and scrolled up in a group called "Purchasing" for almost two minutes before he found the owner's reply: "OK, go with this." That one line was the entire approval record. I asked him which version of the quotation the owner had been looking at. He paused, then said probably the last one.

I'm Johnny Yang, Brightstream Technology, Taiwan. I've led implementation teams for seven years, across 252+ projects in 32+ industries. With purchasing, most companies I've seen assume the whole process is rotten. Lay it out stage by stage, though, and the rot is concentrated in a few places.

Here is the conclusion up front: Purchasing has six stages from start to finish: purchase request, approval, ordering, delivery tracking, receiving, and reconciliation and payment. Small and mid-sized companies rarely have all six broken. Usually only two. The first is approval with no trail. The owner replies "OK" on LINE, and afterward nobody can say which version, which price, or which delivery date he approved. The second is receiving and reconciliation being recorded separately. The warehouse logs incoming goods in its own notebook. Accounting checks the invoice against a different set of data. The two sides never meet until month-end, when the numbers don't match and everyone starts hunting. What these two breaks share is that you can't trace them back afterward. The other four stages hold up even when they're done the crude way, because someone is watching them in real time. So fixing purchasing does not mean putting all six stages into a system. I've seen plenty of companies do that. Most ended up with a heavier process that everyone routes around. What you do is first turn approval into a single versioned record, then force the receiving record and the invoice onto the same order to be matched. Connect those two and you stop roughly 70% of the bleeding. The other four stages can keep running the way they do now until they actually get in the way. If you only have a handful of orders a month, two or three suppliers, and the owner does the buying himself, you should not touch a system yet. The only thing to do is move approvals off LINE.

Lay out the six stages and see what each one hands off

When most owners say "our purchasing is a mess," what's in their head is a vague cloud of anxiety. To fix it you have to cut it apart first, because each stage hands off something different and fails in a different way.

Stage one is the purchase request. It hands off what to buy, how much, and when it's needed. In small companies the request is usually a sheet of paper, an email, or someone walking over to tell purchasing. It looks primitive, but it rarely causes big problems. The person who needs the item is right there, so a mistake gets caught the same day.

Stage two is approval. It hands off consent: who agreed, to what, and what information they were looking at when they agreed. Of the six, it's the shortest, the least noticed, and the most likely to blow up later.

Stage three is ordering. It hands off a commitment. You formally commit to a supplier on quantity, price, and delivery date. Most companies do this stage reasonably well, because the order has to go out, so the format has to look presentable.

Stage four is delivery tracking. It hands off time. It hurts, but it hurts in the moment: sales is chasing, the production line is chasing, purchasing itself is chasing. So even with a whiteboard and phone calls, nothing slips through. It's just exhausting.

Stage five is receiving. It hands off facts: what actually arrived. Is the quantity right, is the spec right, are there defects, does anything go back.

Stage six is reconciliation and payment. It hands off money. The supplier sends a statement or invoice, accounting checks it, then pays.

Put the six side by side and one thing becomes clear. Purchase requests, ordering, and delivery tracking all have someone watching in real time. However crude they are, they won't rot to the bone. Approval, receiving, and reconciliation share a different trait: when they go wrong, you find out much later. Approval, and the receiving-plus-reconciliation pair, are the two places I see break most often on site.

Break one: approval leaves no trail, and you can't trace back what you agreed to

First, the owner's side. Approving on LINE is the right instinct, and there are good reasons for it.

The owner is out on the road. Purchasing sends a picture. Two seconds later he replies "OK," and the order goes out that day. Tell him to log into a system, find a specific order, and press an approve button, and he'll do it the first time. By the third time he's asking, "Can't you just send it to me?" Any process that slows the owner down won't survive three months. I agree with that.

But the problem with approving on LINE isn't speed. It's that he's approving a message, not an order.

What's the difference? Messages get superseded. You send a quotation, the owner says OK. The next day the supplier says that material went up, so you send another version. The owner is busy and doesn't reply. You call him, and he says fine, go ahead. Three weeks later the goods arrive on terms that differ from version one. Accounting asks who approved this, and all you can find is the first message. This almost never blows up on the spot. It blows up months later, during reconciliation or an audit of the books.

An even more common variant: approvals and discussion share the same group. Quotations, specs, supplier business cards, and where to order lunch today are all piled together, and the owner's OK is buried in the middle. Three months later, finding it takes keyword searches and luck.

Fixing this stage doesn't take a big project. Only one thing matters: the approval has to be tied to something with a version. If the order changes, the earlier approval is automatically void and has to be approved again. It sounds bureaucratic, but it's the core of the whole stage. Without it, an approval record can never say clearly which version it approved.

Where the owner actually presses the button is secondary. On his phone, inside a messaging app, either works, as long as what's left behind is "this person approved version X at this time," not an OK floating in a group chat. The smoothest setup I've seen puts the item, amount, and delivery date right in the notification. The owner glances at it and taps, which is faster than opening the picture himself. The slow setups all drop the owner into an interface where he has to dig three layers deep to find the order.

Break two: receiving and reconciliation are recorded separately, and the fighting starts at month-end

The second break is at the back end, and it costs more than approval.

What does it look like on site? Goods come in. Someone in the warehouse counts them and ticks a box in their own receiving log. That record stays in the warehouse. The invoice or statement arrives at the office, and accounting checks it against the order data purchasing gave her originally. The two sides use different paper, different files, different logic, and nobody in between is responsible for matching them.

Day to day you don't see a problem, because most orders really do arrive complete with the right amount. The trouble comes from the messy orders: split deliveries, a few units short to be made up next time, partial returns, supplier miscalculations. Someone handles each of these at the time, but once handled, it lives only in that person's head. At month-end, accounting sees an amount that doesn't match the order. She asks purchasing. Purchasing says the warehouse received it. She asks the warehouse. The warehouse says one batch was sent back. She asks the supplier. The supplier says a credit note was already issued for that batch, and nobody knows where the credit note was sent.

This reconciliation tug-of-war is the same disease as on the collections side, just in the opposite direction. I covered the receivables side in "Statements and collections". The causes on the purchasing side are almost identical: it's not that nobody is working. It's that two departments each keep their own record of the same event, and there's no place where the records meet.

The fix isn't buying a receiving system. It's one simple rule: the purchase order, the receiving record, and the invoice must be viewable together in one place. When all three match, the order passes automatically. When they don't, it sits there waiting for someone to deal with it.

It sounds obvious, but this is exactly what most companies on site are missing. If the warehouse doesn't even record which purchase order a delivery belongs to when it's received, nobody downstream can recover it. So the first real change in this stage usually isn't implementing anything. It's requiring whoever receives the goods to link them to an order number at the moment of receipt. Do that, and even if you're still reconciling in Excel, it works. Skip it, and however expensive the system you buy, you're just moving the mess into it.

StageWhen it goes wrongTouch it first?
Purchase requestCaught on the spotLeave it for now
ApprovalBlows up months later in the booksFix first
OrderingOn the spotLeave it for now
Delivery trackingOn the spot; exhausting but nothing slipsLeave it for now
ReceivingBlows up at month-end reconciliationFix first
Reconciliation and paymentMonth-end, year-endFix together with receiving

Why not do all six at once

When an owner hears "purchasing needs sorting out," the instinct is to find a purchasing module and move all six stages into it. I understand the thinking. If you're spending the money anyway, you want it done in one go. But among the failed implementations I've seen, this approach fails at a particularly high rate, for concrete reasons.

First, going live on all six means requesters learn a new form, purchasing learns a new interface, the warehouse learns scan-to-receive, and accounting learns a new reconciliation process. You're changing the habits of four groups at once. If one group gets stuck, the whole chain stops, and then everyone says, "Let's keep using the old way until the system is stable." Once that sentence is said, the project is basically over.

Second, stages that weren't broken get broken. Many companies run purchase requests and delivery tracking the crude way and it works fine. Forcing them into a system just adds steps nobody needed. Before, you walked over and said something. Now you open a form, wait for approval, fill in the result. People on site won't protest. They'll quietly keep using the old method and enter the data afterward. What you get is a system whose data is always a step behind.

Third, when the scope of change is that big and something goes wrong, nobody can tell which stage was designed badly.

This isn't unique to purchasing. It's true of any process. Which stage to touch first, and how to decide, I covered more fully in "Which process first". For purchasing, the answer is approval and receiving-plus-reconciliation, because both share one trait: when they go wrong, you find out much later, and by then the evidence is gone. Any stage with that trait deserves to be connected first. The stages where someone yells at you on the spot can wait.

When you should not touch a purchasing system yet

This section may matter more than everything above.

The first case is volume too low. A handful of orders a month, the same two or three suppliers, and the owner does the buying himself. In that setup all six stages live in one person's head, so nothing breaks between them. The only thing you should do is move approvals off LINE to somewhere they'll be kept. Nothing else.

The second case is nobody owns purchasing yet. At some companies whoever is free does the buying. Sales places orders, the plant manager places orders, the owner's wife places orders. Implementing a system in that state just produces a pile of data nobody maintains. If you can't define who is responsible, a system only records the chaos.

The third case is that what's actually hurting you isn't purchasing. At some companies purchasing is messy, but the money is really leaking from inventory: stock bought and left sitting, the same material bought twice, items on the books that can't be found. In that case look at inventory first. Purchasing is just the symptom.

The fourth case is awkward: the owner is the one who won't leave a trail. He likes to settle things on the phone or over dinner because it's fast. In that case, however well you build the system, the approval field will always be empty, and everyone will learn to "do it now, log it later." There's no technical fix for this. It comes down to whether the owner is willing.

The first thing you can do tomorrow

Don't look for a vendor first. Don't call a meeting about whether to implement something. Do two small things. Together they take about half a day.

First: pull last month's five largest purchase orders and answer three questions for each. Who approved it? Which version were they looking at when they approved? Did the order change after approval? The ones you can't answer are evidence that break one is bleeding. Most owners feel a bit uncomfortable after doing this. That's normal.

Second: find one order with a split delivery or a partial return, and put the warehouse's receiving record next to the invoice accounting has. See whether the two documents match on their own, or whether someone has to explain before they line up. If someone has to explain, break two is confirmed too.

After these two, you'll have two very concrete things in hand: a stack of approvals you can't trace back, and a set of numbers that don't match. Take those into any implementation conversation and your questions will be completely different. You won't ask "what features does your purchasing module have." You'll ask "when an order is revised, does the earlier approval automatically become void," "can receiving be forced to link to an order number," and "where do orders that fail three-way matching sit while they wait for someone." Ask those three questions and the vendors' answers will quickly show you who has actually worked on site. If you're not sure which stage to touch first, we offer a free 30-minute diagnostic.

Found this useful? Pass it on
30 minutes · free

Bring the one thing you are stuck on — a free 30-minute diagnostic

You do not need to know what system you want. Tell Johnny Yang about the one thing that keeps blocking you, and in 30 minutes you will know whether to skip it, build it with coaching, or pay for a custom build. No pitch — if we cannot help, you will hear that straight.

Johnny Yang replies himself · limited spots each month · or just book a free 30-minute diagnostic

Johnny Yang (熊董)
Johnny Yang (熊董)
Founder, Brightstream Technology

Serial founder leading a small team that builds systems into an owner's everyday work.

As a founder
7 yrs
Projects delivered
252+
Industries served
32+
About Johnny