Spend Analysis

After the carton tender, Rim Nam Foods' CFO asks a simple question: How much do we spend on packaging a year? She gets three answers. Procurement says THB 27 million, the value of the carton contract. Finance says THB 41 million, the balance on the packaging account in the general ledger. Accounts payable says it paid 23 different packaging suppliers, and nobody can say how much went to each.

Nobody is wrong. Each function counts a different thing. But until everyone agrees on one number, nobody can say whether a saving is real, which supplier deserves the next tender, or how much is being bought outside the contract.

That agreed number is the spend baseline, and spend analysis is how you build it. this is the next step in the RFP design post  and the procurement duties map. It covers what a baseline is, how to build one from the data already in your ERP, and what Rim Nam found when it did. 

Remember we follow Rim Nam Foods, a made-up curry maker outside Bangkok.

From a post I made 6 Years ago:  "The humble Spend analysis, a window into your Businesses Health and Mechanics.

Typically used by the procurement team to find and optimise your spend, drive overall cost improvement and effectiveness.

However, the spend analysis is not only to be used by the procurement team as it provides a window into your overall Business effectiveness.

Cross-Functional sharing of insights will support other functions to review processes, procedures and spending, in order to deliver an overall business improvement strategy."

Spend analysis: From Raw ERP data to an agreed spend baseline in five steps.

What is a Spend Baseline?

CIPS describes spend analysis as a process for analysing an organisation's historical purchasing data, with three aims: establish the true spend in each category, find sourcing opportunities by combining demand and reducing supplier numbers, and cut cost through better compliance.

The baseline is the output. For the last 12 months, it shows three things for every purchase:

1. What was bought, by category.
2. From whom, by supplier, with duplicates merged.
3. By whom, by site, plant or business unit.

Analysts often call this the spend cube, because you can turn it and look from any of the three sides. Add price, volume and contract status to each line, and it becomes the reference everything else is measured against: savings, tender volumes, supplier reviews.

The prize can be large. The CIPS brief puts typical savings from spend analysis at between 2% and 25% of total spend. The same brief names the usual obstacle: poor data quality, cited by more than half the companies in the research it draws on.

Why Does It Matter Right Now?

Procurement teams have less time for cleanup than ever. The Hackett Group's 2026 study expects procurement workload to rise 8% while headcount falls 0.9%. With fewer people, you can't afford to tender the wrong category or chase a saving finance won't recognise.

For Thai companies, the invoice data itself is getting better. E-Tax Invoices are still voluntary, but the government offers a 200% tax deduction on eligible e-Tax Invoice and e-Receipt investments, with an extension to the end of 2027 approved in June 2026. Structured electronic invoices carry supplier tax IDs and line descriptions in a consistent format, which makes the first three steps below much faster.

Step 1: Pull the Right Data

A baseline needs three extracts from the ERP, covering the same 12 months, or 24 if you want to see trends:

1. Accounts payable invoices. What you actually paid, to whom and when.
2. Purchase order lines. What you ordered, at what unit price and quantity.
3. The supplier master. Names, tax IDs, addresses and payment terms.

Then add what sits outside the ERP: corporate card spend, petty cash and anything a second plant buys on its own system. In our ERP work, the spend that's missing from the first extract is rarely small. It's usually a whole site or a whole category, such as freight paid through a forwarder's invoice.

Start from paid invoices, not purchase orders. Not every purchase has a PO, but every purchase gets paid.

Step 2: Clean the Supplier List

The same supplier often sits in the master file three or four times, as an English name, a Thai name, an old name before a merger, and a duplicate someone created to avoid waiting for approval. Each record looks like a small supplier. Together they may be your biggest.

Match on tax ID first, then on bank account, then on name and address. Group subsidiaries under their parent, because a converter with three plants is one negotiation, not three. Record every merge so finance can follow the trail.

Step 3: Classify Every Line

Every invoice line needs a category, and the GL account isn't enough. The ledger shows how finance books a cost, not what was bought. A carton, a label and a pallet can all land in "packaging materials".

You can build your own taxonomy or use a standard one. The most widely used is UNSPSC, which classifies goods and services in four levels: segment, family, class and commodity. For most mid-sized manufacturers I'd keep it to three levels of your own, mapped to UNSPSC where you need to compare with outside data.

Classify by value, largest first. The top lines by value are where the decisions are. We aim to have almost all spend classified and stop chasing the last few thousand small invoices. But remember that 20% of spend in large organisations will still add up 2 a nice saving, never discount the tail spend.

Step 4: Add the Context

Spend on its own tells you where the money goes. Context tells you what to do about it. For each supplier and category, add:

• Whether the spend is on contract, off contract or has no contract at all.
• Unit price over time, so you can see the same item bought at different prices.
• Payment terms, because the same price on 30 days and on 90 days is not the same cost.
• On-time delivery and quality data, where operations and quality track it.

That last point matters. In the duties map post I set out five drivers: cost, quality, service, innovation and continuous improvement. A baseline that only holds cost will steer every decision towards price. Putting delivery and defect data next to spend means the first conversation about a supplier covers more than the invoice total.

Step 5: Read It and Get It Signed Off

Three views tell you most of what you need:

1. Pareto by supplier. BCG describes the typical pattern as about 20% of suppliers accounting for 80% of spend, with the rest, the tail, making up most of the transactions. The head needs category strategies. The tail needs simpler buying channels.
2. Suppliers per category. Ten suppliers for one standard item usually means volume is being split for no reason.
3. Price variance for the same item. The same specification bought at different prices by different sites is the quickest saving you'll find.

Then take it to finance and agree it. The baseline price and volume for each category become the reference for every saving reported over the next year. That fixes the disagreement about how savings are counted, which the duties map post showed is one of the most common sources of friction between procurement and finance. Sign it, date it and refresh it every year.

What Did Rim Nam's Baseline Show?

The 23 packaging suppliers turned out to be 17 once duplicates were merged. Two were the same box converter under its English and Thai names. The THB 41 million broke down like this (illustrative figures):

Two things stood out. First, the second plant was buying the same carton specification from local converters at around 14% above the contract price, because nobody had told them the contract existed. Second, eleven suppliers for labels and film was a tender waiting to happen.

Neither showed up in the procurement number or the finance number. Both showed up the moment the data was joined.

Back to the CFO's question. The company that answers it with three different numbers argues about which one is right. The company with a signed baseline answers THB 41 million, a third of it off contract, and has its next two tenders picked before the meeting ends.

I'm not good with Graphic Design, all the images are AI generated.