The State of Category Management in Procurement

A category strategy is worth nothing until it has authority. AI just made that impossible to ignore.

Ask a CPO whether their organisation does category management and the answer is almost always yes. Ask a category manager whether last year’s strategy changed a single decision, and the answer is usually a pause.

That pause is the subject of this piece. The discipline spends its energy on the wrong question. Better frameworks, cleaner data, more AI, higher maturity: all of it sits upstream of the one thing that decides whether a category strategy is real.

This is a guest post written by Lubos Libiak. Twenty-five years in procurement: buyer to head of procurement, consulting, and category management software. Lubos is currently VP of Sales at akirolabs, a software company focused on category management.

Note: This is not a sponsored post. We merely saw this posted as an article on LinkedIn and thought it was a fantastic overview of where we are, and what needs to happen, to give category management as a concept some real teeth.

Read this as a manual of how to turn theatre and perceived self-importance into real influence. Because most CPOs still desperately need this.

A category strategy has no value until it has authority

Not quality, not completeness, not analytical rigour.

Authority: decision rights standing behind it, and evidence that it changed a decision someone would otherwise have made differently. A strategy can be excellent and completely inert, because excellence was never the thing in short supply.

Say it plainly, because anyone who has spent years on both sides of the procurement table has watched it happen: a great deal of category-strategy work is elaborate preparation for decisions the category manager was never allowed to make.

Not all of it. Every organisation has a category or two where it genuinely works. Usually that is because one person spent years earning the standing to make it work, which is itself the point.

But as the default case it holds. The analysis is real, the slides are real, the eight-week effort is real. What is missing is the line of authority that would let any of it bind a decision.

I have asked versions of this in workshops for years.

Put a hundred category managers in a room and ask how much time they spent with their suppliers last year, how many innovation projects sit in their pipeline, how many sustainability ideas they have started in their own category, and you will not get ten hands.

Not because they are poor at their jobs, but because none of it was ever the thing they were held to.

Ask what changed as a result of last year’s strategy and the honest answer is often nothing that would not have happened anyway.

The strategy did not lose to a better strategy. It lost to a sourcing event, a stakeholder with a standing supplier relationship, a budget cycle, a decision made in a corridor. It was never in the room where the decision happened.

Authority is zero-sum. It is not built by the people who need it, it is granted by the people who already hold it.

There is a reason it is so rarely handed over, and it is not oversight. A category strategy gains authority only when someone who currently makes that call loses the right to make it their way: the plant manager with a supplier he trusts, the budget holder who has always chosen her own vendor, the engineering lead who specified the incumbent into the design.

Every unit of authority a strategy acquires is a unit somebody else gives up, and to them it is not an abstraction about operating models. It is a real loss of control over something they are accountable for.

None of them is being obstructive. Each is protecting an outcome they will personally answer for, which is exactly what you would want from them. That is why the transfer is hard rather than merely neglected.

This explains why capability programmes so consistently fail to produce it. Training, frameworks, and maturity models are all ways of improving the strategy. None of them transfers a decision right, because a decision right is not a skill. It is taken from someone. An organisation can run a five-year excellence programme, lift every score on the maturity model, and end with category managers who are measurably better at producing strategies that still decide nothing. At no point did anyone move a decision from one desk to another.

Which puts the question somewhere less comfortable than the usual diagnosis.

If authority has to be taken from someone, the constraint is not the category manager’s capability and it is not the framework. It is whether leadership is willing to spend the political capital to move a decision right, and to hold that position the first time a senior stakeholder objects.

Most do not, and it is rational not to: the cost is immediate and personal, the benefit is diffuse and arrives later. So the mandate stays rhetorical, the strategies keep getting produced, and everyone continues to describe the function as strategic.

There is a fair objection to all of this, and it deserves a straight answer.

No sensible CPO wants a category manager overruling an engineering lead on a specification, or telling a plant manager which supplier he must live with at three in the morning when a line goes down. Those people carry the accountability, and handing procurement a veto over them would be bad governance, not better category management.

But that is not what authority means here, and the distinction is the whole thing. Authority is not the right to decide alone. It is the right to be in the decision. Consulted before the specification is frozen, not after. The trade-off the strategy identified actually weighed. Any departure from it made openly by someone who owns it, rather than happening quietly by default.

The engineering lead still owns the call. What changes is that the call gets made with the category strategy in the room instead of around it. Most category strategies are not being overruled. They are never consulted, and the distance between those two is the subject of this paper.

What the Benchmarks Actually Say

A decade of spend, and the list has not moved

The most interesting thing about the procurement priority stack is not its contents. It is that it has barely moved in ten years.

• Cost control has led for as long as these surveys have been run.
• Resilience has been rising since at least 2020. ESG and regulatory compliance have been climbing for years.
• Talent development is a permanent fixture.

Read one year’s report and it looks like a snapshot of what procurement leadership cares about.

Read ten and something else appears: a list that has been leading and rising and climbing in the same order, through the largest wave of procurement transformation spend in the discipline’s history.

A list this stable is not a description of priorities. It is a measurement of aspiration that keeps failing to become behaviour. There is an innocent reading, that cost simply is the right first priority and the stability reflects sound judgment, and for cost alone that holds.

It does not explain the rest of the list.

Resilience cannot credibly rise for six consecutive years without ever arriving. If it rose, something below it must have fallen. Nothing did. So the rise was a statement of intent rather than a change in behaviour. That leaves two possibilities: the interventions of the last decade did not work, which is a serious thing to sit with, or the surveys record what leaders believe should matter rather than what their organisations actually do, which is the same said-versus-actual gap this paper is about, showing up in the industry’s own data about itself.

65% of CPOs still name cost control their top priority, the same position it has held for as long as these surveys have been run, through a decade of transformation spend intended to broaden the function.

Source: PwC Global Procurement Survey, c. 1,000 respondents*

2–5x more strategic value reported by the small group who did escape it, and what separates them is not methodology or spend. It is decision logic and governance: the conditions that let a strategy decide something.

Source: Kearney Assessment of Excellence in Procurement

Frameworks are not the problem; the portfolio models remain broadly sound. What is consistently absent is the set of conditions that let a strategy decide anything: governance clarity, role design, decision rights, and a leadership mandate that holds when it is tested.

The Hackett Group’s 2025 research arrives at the same conclusion. Category management is among the most widely adopted capabilities, while AI shows high investment intent and mostly pilot-stage impact. Adoption is not the constraint.

It is worth being straight about what this evidence is.

It is associative and self-reported, drawn from organisations grading their own strategic value. This does not prove that authority causes performance. The narrower claim holds. Across these benchmarks, the functions reporting the most value are not the ones with the best methodology or the largest technology budget. They are the ones describing clear decision rights, and governance that produces commitments. That is a pattern, not a proof.

The falsifiable version is available to any organisation willing to run it on itself. If authority is not the constraint, the categories where strategies change decisions should look no different from the categories where they do not. I have not yet worked with an organisation where that held.

Both readings of the stalled list point the same way. If a decade of spend has not changed what procurement prioritises, then the constraint is not knowing what to prioritise. Everyone has known for ten years.

The constraint is that resilience, innovation, and sustainability have never had what cost has: a number someone will sign for.

Excellence is not scarce. Authority is.

The Authority Gap

Three accounts of one function, all accurate, all incompatible

The same procurement function looks completely different depending on where you stand in it. None of these three people is lying. Each is describing their real experience of the same system.

The gap: The distance between how leadership believes the function operates and what the category manager actually lives is the authority gap. It is remarkably consistent across organisations, sectors, and maturity levels, and it does not close with a better template. It closes only when a strategy carries decision rights and someone can point to a decision it changed.

The test: It has nothing to do with how polished the document is. Go to the moment a real decision was made in that category, and ask whether the strategy was in the room.

Without the strategy in the room — A category strategy is produced, but the decision is made without it. Nothing would have changed: remove the strategy and the outcome is identical. It was never overruled. It was never consulted.

With the strategy in the room — A category strategy is produced, and the decision is made with it in the room. Someone chose differently and can say what the strategy changed. It was in the decision, not around it. That is what authority looks like.

How the Gap Shows Up

The authority gap has a signature

You do not diagnose a missing mandate by asking about it. You recognise it by the patterns it leaves behind, and once you know them, they are hard to unsee.

1. Strategy becomes theatre

Strategies get produced to show category management is happening. Comprehensive, well-structured, and never cited in an actual decision. It is the authority gap in its purest form: full production, zero authority.

2. Governance produces meetings, not decisions

Review forums fill calendars at every level while decision rights stay undefined. Nobody is sure what a category manager may commit to, so the forums generate updates instead of choices. Activity gets mistaken for authority.

3. Initiatives accumulate because nobody can close one

The average category manager carries more initiatives than can be run well, because stopping one requires a decision, and deciding is the authority nobody has. The list grows because pruning it is harder than adding to it.

4. The real decision happens in a corridor

The strategy says one thing; the choice gets made by a budget owner, an engineering lead, or a stakeholder with a standing supplier relationship, outside the process the strategy governs. Procurement is informed, not consulted.

5. The strategy refreshes on the calendar, not on a decision

It gets updated when someone asks for a new deck, not when a real choice forces it to change. The version I have seen most often is a strategy built over weeks, presented once, then left in a drawer for eleven months until the next refresh, at which point the team quietly concedes they worked on one or two items and never had time for the rest. A document that moves once a year is not driving the decisions that happen the other fifty weeks.

Why Procurement Stays on Savings

Savings is the only value with a signature behind it

Everyone agrees procurement should broaden beyond cost. Everyone has agreed on this for around 25 years, and it has barely moved.

The reason is not cultural. It is structural.

Savings is the one form of value in procurement with decision rights already attached. It has:

• An agreed baseline
• An accepted unit
• And a CFO who will sign for the number.

It is tempting to say the others have none of that, and that is too easy.

• Sustainability has a baseline and a unit, in some organisations a better one than savings: tonnes of CO2 equivalent, scoped, audited, and increasingly reported under regulation.
• Resilience can be measured too, in disruptions per year, recovery time, qualified alternate sources.

The measurement problem is real but it is not the binding one.

What savings has, and the others do not, is a price.

A euro saved converts directly into the number the business is actually run on, so a category manager can carry the trade-off into a room and have it weighed.

Nobody can tell you what 200 tonnes of CO2 is worth against a 4% price increase, or what one avoided disruption is worth against a second source costing 8% more, because nobody has ever had to sign for that conversion.

I have sat in rooms where that question was put directly and watched it go unanswered, not because the people were unserious but because no answer existed to give. The unit exists. The price does not. So the strategy can argue for resilience, but it cannot price it.

An argument that cannot be priced loses to one that can, every time. Put a category manager in front of a strategy with authority over cost and none over anything else, and they will retreat to cost. Not from a failure of ambition, but because it is the only place their work can actually decide something.

Building the missing scoreboard is harder than saying it, and the usual attempts fail in a recognisable way.

A resilience measure that counts dual-sourced suppliers gets gamed within two quarters by naming a second source nobody has qualified. The measures that survive share the properties that make savings work, and measurement is only the first of them. A baseline agreed before the work starts, not reconstructed after it. A unit that cannot be improved without changing something real. A stated price at which that unit trades against cost. A named person accountable for the number moving.

The first two are the ones organisations attempt. The third is the one almost nobody sets, and without it the trade-off never reaches a room where it can be decided. Most organisations attempt the metric and skip the signatory, which is precisely why the new measure never acquires authority either, and why the function slides back to the one number that has always had all three.

This settles a long-running argument. The Kraljic matrix, the demand-supply power read, the purchasing chessboard: these are diagnostic tools for assessing a category and its suppliers. No competent category manager treats them as instructions and follows the output blindly. They were never the cause of procurement’s narrowness.

The methodology was never the constraint. A strategy with no authority and no scoreboard was.

The Most Expensive Proof

Even the best-paid strategy drains into sourcing

The clearest evidence that authority, not production, is the scarce thing comes from the part of the industry that has removed the production constraint entirely.

Large enterprises routinely bring in the top consulting firms to build the category-management capability itself: the operating model, the templates, the governance, and in many cases the category strategies that populate it. The work is genuine and often excellent.

But follow those strategies over the life of an engagement and a pattern emerges. They resolve, almost every time, into sourcing events. The reason is worth stating precisely, because the pricing differs by firm. Among the procurement specialists, the houses whose entire model is procurement, a meaningful share of the fee is commonly tied to secured savings through gain-share or contingency.

The strategy firms price differently, usually fixed fee or retainer, but the engagement is still judged on the savings number in the closing report, and the next engagement depends on it. The mechanism differs; the scoreboard does not. The strategy gets produced, the engagement is measured at the sourcing event, so that is where the effort flows.

Even the best-resourced operators in procurement, with their own remuneration on the line, produce a strategy and then let it drain into the sourcing next to it, because sourcing is the only place the value can be booked. It is the same gravity that pulls the internal category manager toward savings, with one difference: here it is written into the contract.

It also does not stop at sourcing events. A great deal of the value gets booked by reopening what is already signed: renegotiating with incumbents, pressing hard on a supplier who has been in place for years, converting a relationship into a percentage. The leverage is real and the tactic works.

What it draws down is an asset nobody has ever put on a balance sheet. The supplier who absorbed a late specification change without a claim, who expedited without invoicing for it, who gave you first call on capacity in a shortage, was providing value continuously and invisibly. Squeeze that relationship hard enough and the 3% appears in the report. The thing that was spent to get it appears nowhere.

And when those operators go hard in a negotiation, leaning on an incumbent, pressing with real leverage, they are not drawing on their own authority. They have been lent it, granted the mandate to speak for the organisation for the duration of the deal. When the engagement closes, the deck remains and the authority leaves with them.

The category manager who inherits the strategy is asked to drive it on a level of access and standing they were never given. It does not fail because it was poorly thought through. Its authority was rented, and the lease is up.

What I have learned over 25 years is that the framing itself is wrong. The choice is usually presented as how hard to squeeze, which assumes the value is a fixed quantity and the only question is who ends up holding it. Often it is not fixed. The alternative to squeezing a supplier is not conceding to one. It is listening to one.

A supplier asked where the cost actually sits will usually tell you, and the answer is usually not the price. It is a tolerance nobody has questioned in nine years, a packaging format designed for a line that was replaced, a specification written around a constraint that stopped existing three product generations ago. Those conversations create value rather than move it, which is why the supplier engages at all. They also routinely produce more than the negotiation would have. They require the buyer to treat the supplier as a source of engineering judgement rather than a counterparty to be managed. That is a different posture entirely, and it cannot be faked for the duration of a workshop.

Then the problem returns in its original form. To act on any of it, the category manager has to be inside the decision that sets the specification, and that is exactly the decision they were never granted. So the supplier’s proposal goes into a report. The specification stays as it is. The negotiation happens anyway, because the negotiation is the only part of the conversation they have the authority to run. The relationship gets spent on a percentage while the larger value sits one decision upstream, visible to everyone, actionable by nobody in the room.

This happens to consultants and employees alike, and it is not a failure of imagination. Authority is not only event-shaped. The event recognises exactly one kind of value.

There is a reading of this that defends it, and it deserves an answer rather than a dismissal. Decisions need decisive moments. The sourcing event is where the commercial lever actually exists, where a supplier can be changed and a price can be moved. So perhaps authority belongs exactly there, and the strategy’s real job is to prepare those moments well rather than to hold standing power of its own.

The difficulty is what has already been settled by the time the event arrives.

• The specification is frozen.
• The supplier list is inherited.
• The timing is set by a contract expiry nobody chose.

Authority over a decision space that somebody else defined is not strategic authority, it is permission to optimise inside constraints already fixed. That is why the same categories produce the same savings from the same suppliers year after year while the strategy documents keep changing.

The authority at the event is real. It simply arrives too late to change what the event is about.

Authority is event-shaped. It switches on for the sourcing moment and off for the strategy, for consultants and employees alike.

The pattern underneath everything else

The discipline is organised around the few moments where authority is granted. Category strategy is the work where it is withheld.

 

What AI Changed

The strategy is now easy to produce, and it proves nothing

For as long as a category strategy was slow and expensive to produce, its existence carried a quiet second meaning: someone had spent weeks on it, so the thinking must have happened. The document was weak evidence of rigour, and organisations leaned on that proxy more than they admitted.

AI removes the proxy. Anyone can now generate the output of a category strategy: the sections, the credible language, the confident tone. For almost any category, in a couple of hours. Absent real grounding it will be generically useless, but that is not the point. The point is that the finished document no longer proves anything. When a strategy deck was costly, its existence signalled effort.

Now that it is free, it signals nothing.

This is why AI raises the stakes rather than resolving them: it removes the last hiding place for strategy as theatre. Now that the document is free, the only thing that separates a real strategy from a generated one is whether it changed a decision, which is to say whether it had authority.

AI can produce the thing that looks like the work. It cannot produce the thing that is the work. Decision rights, mandate, and evidence of influence are organisational facts, and they are exactly what no model can manufacture.

I watched this before I had language for it. In a workshop with an organisation of around a hundred billion in revenue, we rebuilt category strategies that had taken their team nine months, and did it in roughly five hours across two days, to a standard that went to their CPO as final output.

Everyone could see the time saved, but nobody could answer whether any of it would change a decision. Their CPO said the truest thing I heard that year: we are learning to walk, and you are showing us how to run.

The augment-versus-automate distinction is worth a closer look. AI is genuinely useful to a category manager who has authority: it compresses the synthesis, reads the annual report, surfaces the supplier asking for a 3% increase while posting record profits, and hands back hours. It does very little for the one whose value was the production of the strategy document, because that value has just gone to zero.

The adoption numbers say it quietly. I have seen deployments with 300 people who should have been in the system and 17 who logged in that month, and the tool was not the problem. Content without context is nothing, and context is the same raw material authority is built from.

 

AI made the deck free. It made authority the only thing worth paying for.

What Good Looks Like

A function with authority is recognisable on sight

The organisations that have closed the gap do not share a methodology or a maturity score. They share a set of markers, and every one is about authority rather than analysis.

1. Strategies force a choice

Short, explicit about the trade-off, and written to change a decision. If pulling the strategy would not change what the organisation does, it does not get written at all. The cost: the category manager gives up the safety of a document broad enough that nobody can disagree with it.

2. Decision rights are written down and known

Everyone can say what a category manager commits without asking, and two leaders asked separately give the same answer. The cost: the ambiguity is gone, and ambiguity was useful to whoever had been deciding in its shadow.

3. Governance decides and moves on

Forums resolve trade-offs and exceptions rather than reviewing status. Senior leaders attend for the decisions, not the updates. The cost: leadership can no longer attend without deciding, which is more demanding than being briefed.

4. Overrides are examined, not absorbed

When a stakeholder goes around a strategy, it is surfaced and discussed, because that is the moment the mandate is either real or revealed as rhetorical. The cost: somebody senior has to be told, in front of others, that they went around the process. This is the marker organisations fail most often, and the one that decides whether any of the others are real.

5. Value beyond savings has a scoreboard

Not measured, priced. Someone has stated what a tonne of CO2 or an avoided disruption is worth against cost and put their name to it, so the trade-off gets settled in a room instead of deferred. The cost: an executive has to commit to a number they would rather leave open.

What Has to Be Built

None of this makes category management cheap

It makes the distinction between a strategy and its authority impossible to ignore. Producing the strategy document was never the expensive part. It only looked that way because it was the visible part, and for years the most visible version of it carried a seven-figure invoice.

The expensive part, the part that was always the actual work, is everything that gives a strategy authority: the decision rights, the alignment secured before the decision rather than after, the governance that makes a choice bind, the evidence that it changed what the organisation did.

AI drives the cost of the document toward zero and leaves the cost of authority exactly where it was. If anything it raises it, because once the document proves nothing, authority is the only thing left to build, and it is the harder thing to build by a wide margin.

The conversation worth having is not about the next framework or the next tool. It is about where authority actually sits.

None of the five markers above comes from a maturity model, and none of them needs a consultant.

Walk your own function past them honestly and count how many it can actually claim. Start with the first: name one decision this year that went differently because a category strategy existed, not a sourcing event the strategy described afterward, a decision that changed. If the room goes quiet, that is the finding, whatever the maturity score says.

The uncomfortable part is that the fix does not live in procurement.

Every one of those markers is set by leadership behaviour, not by category managers working harder or a better framework. That is bad news for anyone hoping to buy their way out of it, and good news for anyone willing to move a single decision right and defend it the first time it is tested. It is a smaller step than a transformation programme and it changes more.

If you want somewhere to start, do not start with the largest category. Start with a decision that is already contested, already recurring, and already has an owner who will notice it moved.

• Contested, because a decision nobody disputes proves nothing when it goes your way.
• Recurring, because a one-off establishes no precedent and will be treated as an exception.
• And owned by someone who will notice, because the first defended override is the moment the mandate becomes real to everyone watching.

The first one quietly conceded is the moment everyone learns it never was. That is one decision, in one category. It will tell you more about your function than any maturity assessment.

If you run that exercise and the count is lower than you expected, I would be interested to hear it. The pattern is consistent enough that I no longer find it surprising, but the specifics differ everywhere, and the organisations that get this right did not do it by adopting somebody else’s model.