Spend visibility is a start. Accuracy is the goal

Seeing spend early matters. It helps organisations stay in control before money is committed. But early visibility only helps if the information is accurate. If it is based on guesses, the problems simply show up later in the process.

When people do not know which supplier to use, what something should cost, or how to describe it, they fill in the gaps themselves. That affects everything that follows. In many cases, the true picture only becomes clear when the invoice is reviewed.

That is why guided buying matters. It helps buyers choose the right supplier, see the right items and prices, and raise orders using accurate information instead of free text and assumptions.

Most organisations think they have a spend control problem.

The CFO sees spend running over budget, the Head of Procurement sees maverick spend impacting the procurement team with new supplier setups and the AP Team is spending countless hours managing suppliers chasing payments while internal functions wrestle with resolving mis-matches between invoices and orders.

These are real problems. They are not the root problem.

The root problem is that the data the organisation is trying to control was already wrong before it reached any of these people. By the time an issue surfaces in finance, the fix is correcting symptoms, not causes.

This article is about where the cause actually lives - at the requisition stage - and what changes when you fix it there.

It is like buying from Argos without the catalogue.

If you used Argos in the 1990s, you remember the routine. Walk in, pick up the catalogue, find what you wanted, write the item code on a slip, hand it in. Total time: ninety seconds. The cashier read the slip, the back office had what you ordered, and what came out matched what you wrote down.

Now picture the same shop without the catalogue. You would guess item names. You would guess at prices. You would invent a description that sounded right. The cashier would walk to the back to ask a colleague. Inevitably all of the slips would have to be corrected and the data the till captured at the counter would not match what left the shop.

That is exactly what most organisations are asking employees to do every day at the point of purchase. We call it manual requisitioning, or free text requisitioning. The label is not the point. The practical effect is the same: people fill in a blank slip without the catalogue.

They might be ordering A4 paper. The organisation may already have ten stationery suppliers on file, but none are clearly signposted to the buyer, so they pick whichever name they recognise or invent a new one. They do not know the agreed price. They do not know the right code. So they type what seems reasonable - £5, a supplier name they recall, a description that will be clear enough later.

The request is submitted. But the information contained in that request, whilst done with the best of intent, is unhelpful.

If the information starts wrong, and stays wrong 

The actual price might be £2 or £10. The supplier might be one procurement specifically negotiated away from. The category might be assigned to the wrong department. None of that is deliberate. All of it affects what the organisation thinks it is spending.

Manual purchasing is usually discussed as an accounts payable problem because the symptoms surface later - invoices that do not match, codes that need correcting, finance teams intervening by hand. Those are downstream effects, not the root cause.

The root cause is that the organisation is capturing spend intent through estimates rather than accurate, structured data.

If employees are guessing values to keep work moving, then visible spend is not really visible at all. It just looks like control.

A mistake that compounds 

There is one error in particular that deserves its own attention, because of how quietly it does damage: cost-centre coding.

When an employee raises a draft PO, the code applied at that moment - the department, the cost centre, the category - usually flows all the way through to payment without being reviewed again. The reason is structural: coding is decided at the front of the process, and the people downstream have neither the context nor the time to question it. They are matching invoices, not auditing intent.

For example, If the order to assigned to the Marketing cost centre when it was actually an Operations purchase, that error becomes part of the official record at the moment the purchase order is raised. So even if the invoice matches the PO. The PO matches the purchase. Everything looks fine. However, the attribution is wrong, and it becomes harder to spot and significantly harder to unpick later on.

This isn’t just a spreadsheet headache, real decisions that flow from these numbers - budget reviews, hiring approvals, project cuts - are all now being made on data that isn’t accurate.

This is the silent way budgets get distorted. Not by anyone gaming the system, but by a coding decision made in twelve seconds by someone who isn’t being guided to the right answer.

Guesswork can be worse than no visibility 

At first glance, a manual purchasing looks better than an off-system purchase because at least something has been recorded. But if the supplier, price and code are all guesses, the organisation is making decisions on data that appears complete without actually being accurate.

That can be more damaging than an obvious gap because false confidence is hard to detect. Forecasts look informed. Reports look populated. Approval workflows look compliant. The underlying picture of demand, supplier use, and committed spend is still wrong.

This is why the problem cannot simply be solved by asking users to be more careful. Nobody can be expected to remember approved suppliers, negotiated prices, and coding structures from memory. If the system depends on them knowing all of that, it’s the system is the problem.

The solution is not to train users harder - it is to make the right route the easiest one.

Guided buying fixes it at the source 

The answer is not more manual effort. It is guided buying.

Guided buying changes the point at which control happens. Instead of asking users to invent or estimate purchasing data, it presents approved suppliers, contracted prices, and structured item information inside the buying journey itself. The catalogue, in other words, is back on the shelf.

The user ordering A4 paper no longer has to guess whether the right price is £2, £5 or £10. They select from what has already been approved. The supplier is correct. The price is the negotiated one. The category code is applied automatically. The requisition enters the process with information finance and procurement can trust.

With Proactis Marketplace, users search from pre-approved suppliers, see negotiated pricing and submit requests with structured data ready to flow into Unit4 Financials by Coda (Unit4 FbC).

The result is not just a better user experience or fewer AP exceptions. It is a more accurate picture of committed spend from the moment demand is created.

Procurement value is only realised when negotiated agreements show up in actual buying behaviour. If the requisition stage is still based on guesses, that value leaks away before the purchase order is even created.

Why this matters more in an AI era 

There is one more reason to fix the requisition stage now, and it has nothing to do with today's AP team.

In our upcoming thought leadership article on AI in Source to Pay, we discuss that AI is only as good as the data feeding it. The most accomplished chef in the world cannot produce a clean meal from spoiled stock, mislabelled cuts and a half-empty larder. AI in procurement and finance is the same. It learns from what your processes captured, then makes decisions or recommendations on the basis of that learning.

If today's requisitions are full of guessed suppliers, placeholder prices and wrong department codes, the AI you deploy in two years' time will treat that mess as normal. It will recommend the wrong suppliers. It will spot the wrong anomalies. It will miss the right ones. Faster and at scale, but still wrong.

Fixing the requisition stage is not just an AP improvement. It is foundation work for the AI you want to use next. The organisations that get genuine value from AI in three years are the ones building disciplined, accurate spend data today.

The better question to ask

If teams are still manually correcting purchase order data, the question is not only why invoices are failing to match. The better question is whether the original requisition ever reflected the real purchase in the first place.

That is why guided buying matters. It does not just solve an administrative symptom. It addresses the root problem by replacing guesswork with accurate, governed purchasing data from the start.

Visibility is a start. Accuracy is the goal.