Skip to content
All articles
Guides

Who Should Own Your Spend Taxonomy?

A spend taxonomy decays the moment nobody owns it. Here is who I think should own it, the naming rules that keep it readable, and how to change it without starting over.

By Aditya Chavali

Every spend taxonomy starts clean.

A team spends a quarter on it. Or a consultant hands it over in a binder. On day one, it makes sense.

Two years later it has "Other Services", "Services Other" and "Misc Services". Four levels down there is a category called "Laptops Dell 14 Inch". Nobody remembers who added it. Nobody wants to be the one who removes it.

The taxonomy did not fail. Ownership did.

This question came up in nine of my conversations with customers, prospects and partners this fall. Not which taxonomy to use. Who owns it once it exists.

What is a spend taxonomy?

A spend taxonomy is the structure you use to group what your organization buys. It usually runs from broad to specific. Level 1 might be "IT". Level 2 might be "Hardware". Level 3 might be "Laptops".

Every invoice line and purchase order line gets a place in that structure. That is what spend classification does. The taxonomy is the map. Classification puts each line on it.

A good taxonomy is a model of how you buy. It reflects markets and suppliers, not departments or GL codes. When it works, a category manager can open it and see their world.

It belongs to the people who use it

Should IT own the taxonomy? Finance? The analyst who built the reports?

My answer is procurement. Specifically, the category managers who make decisions with it, with one analyst as the steward who keeps it consistent.

The reason is simple. The people who buy are the people who notice when the structure stops matching reality. A new kind of spend shows up. A supplier changes what it sells. Two categories turn out to be one. Category managers see that first.

IT owns the systems. Finance owns the chart of accounts. Those are different jobs. Mixing them is how a taxonomy ends up mirroring GL codes instead of markets.

The steward matters as much as the owners. Category managers propose changes. The steward makes sure a change in one place does not break the logic somewhere else. One person holds the pen. Many people hold opinions.

Corporate wants one. Divisions want their own.

This is the tension I hear most.

A division runs its own buying. Its category managers want a structure that fits their parts and their suppliers. Corporate sourcing wants one view across every division, so it can see the whole relationship with a supplier.

Both are right.

The usual compromise is to force one taxonomy on everyone. Then the divisions keep their real structure in a spreadsheet on the side. Now there are two taxonomies, and only one of them is visible.

A better answer is to agree on the top levels and let the lower levels flex. Level 1 and Level 2 are shared. They are how corporate rolls things up. Below that, a division can go deeper where its buying is more specific.

SpendCraft is taxonomy agnostic. It can hold more than one taxonomy over the same spend. Corporate can keep its view. A division can keep its own. The underlying lines stay the same.

Naming rules that hold up

Most taxonomy problems are naming problems. These are the rules I see teams settle on again and again.

Do not repeat the parent in the child. If the parent is "IT Hardware", the child is "Laptops", not "IT Hardware Laptops". The path already says where you are.

Add a qualifier only when a name repeats. "Maintenance" under Facilities and "Maintenance" under Fleet are different things. Qualify both. A name that appears once does not get a qualifier.

Attributes are not levels. Brand, size, region and contract status are attributes. They belong in a detail field, not in a new level of the tree. Add "Dell" as a level and "Lenovo" follows. Then the tree is about brands, not markets.

Use the words your buyers use. If every category manager says "MRO", do not name the node "Maintenance, Repair and Operations Supplies". Short names get used. Long names get worked around.

Keep "Other" small and visible. Every taxonomy needs a place for what does not fit yet. When it grows, that is a signal to add a category. It is not a place to hide spend.

None of these rules are clever. They are just easy to forget when ten people edit the same tree over two years. That is why the steward exists.

Change it without starting over

The old reason taxonomies went stale was cost.

Changing the structure meant reclassifying everything. Reclassifying meant a ticket, or a wait for the month end refresh, or a new project. So teams stopped asking. The taxonomy froze, and reality moved on without it.

That cost does not have to exist anymore.

In SpendCraft, the steward can change the structure and reclassify on demand, with no new implementation. Classification Review keeps a person in the loop on what moved. Proprietary AI does the classifying. People decide whether the result is right.

So small changes can happen often. Merge two categories this month. Split one next quarter. The taxonomy keeps up with how you buy, instead of being rebuilt every few years.

Make ownership visible

Ownership that lives in an org chart fades. Ownership that shows up in the work holds.

A category manager can keep their owned categories in a Space, with findings and Caddie work in one place. When the taxonomy is right, that Space is right. When it drifts, they are the first to notice, because their own Space looks wrong.

The steward can watch the Data health Space. It shows classification coverage, vendor cleanliness and data freshness. A drop in coverage is often the first sign that new spend has arrived with no clear home.

A Scout attached to that Space can watch spend in the "Other" category. When spend there grows, it reports the change. That is the signal to add a node, before a quarter of new spend disappears into a bucket nobody reads.

Clean vendor names help too. A taxonomy cannot do its job if one supplier sits under five names. Vendor normalization and classification work as a pair.

The real test

Ask a category manager to locate their own spend in the taxonomy without help.

If they can, it is working.

If they need the analyst, or an export, or a meeting, it has drifted away from the people it is for.

Fix the ownership first. The structure follows.

For more on how classification, normalization and analysis fit together, see our spend analysis guide.

The Savings OS.

See it on your own data.

A Proof of Value works one category or one moment end to end.