The Future of Procurement: From Cost Centre to Growth Engine

If you buy things for a living, you have heard this speech. Procurement is becoming strategic. Procurement is getting a seat at the table. It has been said at every conference since about 2005.

Mostly it has not come true. And the reason is simple. Nothing forced it to.

Something is forcing it now. Not a new idea. A sum that no longer works.

#First, what procurement actually is

Skip this part if you work in procurement. It is here because a lot of people who read articles like this one are finance leads, founders or operations managers who inherited the job.

Procurement is the work of buying everything a company needs to run, and managing the companies you buy it from. Not the products you sell. Everything else. Software, freight, packaging, cleaning, laptops, temporary staff, spare parts, consultants, electricity.

In a well-run company that work has five parts:

Deciding what to buy and from whom

Working out what the business actually needs, finding suppliers who can provide it, comparing them properly, and choosing. This is usually called sourcing.

Agreeing the deal

Negotiating the price and the terms, then putting it in a contract. What happens if they are late. What happens if you want to leave.

Doing the paperwork

Turning a request into a purchase order, checking the invoice matches, getting it approved, getting it paid. This is the biggest part by volume and the smallest part by value.

Watching the suppliers

Are they delivering on time? Are they charging the agreed price? Are they about to go bust? Is a contract about to renew itself while nobody is looking?

Knowing what you spend

Being able to answer, quickly and correctly, how much the company spent on a category last year and with whom.

In most mid-sized companies, one or two people do all five. That is the whole problem in one sentence.

#Why procurement gets called a cost centre

A cost centre is a part of a business that spends money but does not directly bring money in. Sales brings money in. Legal, IT and procurement cost money. That is the whole idea, and it is not an insult by itself.

It becomes a problem when it turns into how people think about you. Rich Weissman, writing in Supply Chain Dive, remembered the joke his colleagues used to make: "They didn't just fire Fred. They put him in purchasing so he could really be punished."

Two things keep the label stuck.

The first is that procurement is usually judged on one number: savings. One number invites one argument, every year, about whether the number is real.

The second is that procurement is usually called in late. The product is designed. The supplier has been chosen. The date has been promised. Then someone asks procurement to get the price down. By that point price is the only thing left to change.

#What changed in 2026: the sums stopped adding up

The Hackett Group runs an annual study of what procurement teams are planning. Their 2026 research says procurement workload will rise about 8% next year, while both headcount and operating budgets fall.

Reporting on the same study fills in the detail: headcount down about 0.9%, operating budget down about 0.4%. Together that leaves a gap of roughly 8.9% in productivity and 8.4% in efficiency. The plan to close it is a 6.1% increase in technology spending.

Compare that with the year before, when the same study expected work to rise 9.8% and staff to rise 1%. Last year you got more work and slightly more help. This year you get more work and less help.

Now put the savings picture next to it. In the 2026 study, 45% of teams expect their savings to increase. A year earlier it was 55%. Another 35% expect savings to stay flat.

Read those together and you get the real story of 2026. More work. Less money. And the one number your function is judged on is getting harder to grow.

A team in that position has two options. Change what it spends its hours on. Or quietly stop doing the work nobody is watching. In practice the second one happens first, and it is always the same work: supplier data, catalogue upkeep, contract checks, small suppliers. That is how a procurement team becomes a cost centre. Not by having the wrong attitude. By running out of hours.

#Where the hours actually go

Invoice processing is the clearest example, because it is the one part of the job with good public benchmarks.

Ardent Partners measured what it costs companies to process a single supplier invoice, from arrival to payment. Not just the software. Everything, including the people.

The gap between $2.78 and $12.88 is more than four times. For the same task, done to the same standard, in the same year. That gap is not caused by clever software. It is caused by whether anybody owns the process.

And every one of those hours comes out of the same small team you are asking to build supplier strategies.

The other big drain is people buying outside the proper process. The industry calls this maverick spend. It is worth being careful with that phrase. Tim Jones, formerly VP of Business Operations at Epic Games, put it well on the Art of Procurement podcast:

One of the most dangerous phrases in the procurement dictionary is maverick spend. It implies that there are people out there who are ignoring us or somehow doing the wrong thing, when in reality they're just a bunch of users who we've often let down, who are so exhausted by the bureaucracy we've built around them that they found a better way to get the business done.

Tim Jones, former VP Business Operations, Epic Games

That is the right way to read it. When people go around your process, the process is usually slower than the alternative. Fix the process and the behaviour changes on its own.

#The savings problem nobody says out loud

Here is something most procurement teams have felt and few have named. The savings you report and the savings your CFO believes are two different numbers.

It is not usually because anyone is lying. It is because you are counting different things.

Procurement counts savings by category, against the last price paid, or against the next best quote. Finance counts against the budget, by cost centre, in the accounts. Bain's work, reported by CFO.com, describes exactly this mismatch: procurement tracks categories, the business tracks budgets, and the two never line up.

So a saving can be completely real and still be invisible. The team spent the money on something else. The volume went up. The budget was never reduced. Nothing appears in the accounts, so nothing feels true.

There is a second version of this, which is worse, because it involves the work that matters most. The Procurist calls it the savings procurement never gets paid for: stopping a price rise, avoiding a bad contract, catching a supplier problem before it happens. That work never shows up anywhere.

Cost reduction has a ceiling. You can only negotiate so deep before you hit supplier profitability. Cost avoidance has none.

The Procurist

The fix is boring and it works. Agree the starting number with finance before you do the work, not after. A baseline agreed afterwards is a debate. A baseline agreed beforehand is a measurement.

#AI in procurement: what is real and what is not

This is the part of every procurement forecast that gets loudest. So let us be careful, because the honest answer sits between the two loud answers.

Start with what is genuinely happening. In the Hackett 2026 research, 43% of organisations are actively working on AI, nearly double the year before. Eighty percent of procurement executives now call AI the most important change facing the function. Most of them are not building anything: 69% get their AI through features added to software they already own.

Now the other half of the picture.

• In the same Hackett research, only 12% have AI running at large scale.
• In the 2026 ProcureCon CPO report, only 11% of procurement leaders say they are using AI with measurable impact, even though 89% describe themselves as ready for it.
• The year before, Hackett found 49% of procurement teams had piloted generative AI, but only 4% had deployed it at scale.
• Gartner found that 74% of procurement leaders say their own data is not ready for AI.
• MIT's Project NANDA studied 300 public AI deployments and found 95% of pilots produced no measurable effect on profit or loss.
• Gartner expects more than 40% of agentic AI projects to be cancelled by the end of 2027, because of cost, unclear value or weak controls.

Put simply: a lot of people are trying, a few are succeeding, and the difference between them is almost never the software.

Deloitte's 2025 survey of more than 250 chief procurement officers across 40 countries asked what they estimate they get back from generative AI. The average was roughly two times the investment. The leading group said 3.2 times. The rest said 1.6. Worth noting honestly: the same report's summary page quotes the leaders at 2.8 rather than 3.2, so treat the exact figure with care. The pattern is what matters. Same tools, very different results.

Why? Look at what companies actually use AI for. In Deloitte's data the top uses are spend analysis and dashboards (53%), writing RFP and RFQ documents (42%), and summarising contracts (41%).

Every one of those depends on the quality of your records. Spend that is properly sorted into categories. A supplier list without duplicates. Contracts stored somewhere a computer can read them. Point AI at a bad record and it produces confident nonsense, faster than a person could produce careful nonsense.

The clearest statement of this came from a research analyst rather than a vendor. Saurabh Gupta, Chief Strategy Officer at HFS Research, described what came back when his firm asked around a hundred C-level executives what was blocking them:

The two biggest barriers were culture, and two was mindset. Nobody said anything else. Probably the third was talent, lack of talent. None of these has to do anything with technology.

Saurabh Gupta, Chief Strategy Officer, HFS Research, on the Art of Procurement podcast

The same conversation contains the line every buyer of procurement software should read twice: there is no point making something that does not work more efficient.

#What the strong teams actually do differently

Two large studies compare the best procurement teams against everyone else. They were done by different firms using different methods, and they broadly agree.

The Hackett Group compared what it calls Digital World Class procurement teams with typical ones in July 2025:

Deloitte's version of the same comparison looks at whether teams hit their plans:

Look at the last row. On savings the gap is 16 points. On helping the business innovate it is 32 points. The most strategic thing on the list is the thing the weaker half almost never manages.

That is not because they lack ambition. It is because they are busy. In Hackett's numbers the strong teams spend 26% more of their time analysing information instead of collecting it. That is the whole difference, expressed as a percentage.

There is also a wider signal. McKinsey's benchmarking, reported by Procurement Magazine and Supply Chain Management Review, links top-quartile procurement maturity to EBITDA margins at least five percentage points higher than less mature peers. Treat that as a correlation, not a promise. Good companies tend to be good at several things at once.

#The four kinds of value, in plain terms

The weak version of this argument says procurement should be valued for things that cannot be measured. That argument loses every budget meeting it enters.

The strong version is that procurement already creates four kinds of value. Most teams only report one.

1. Money saved

Lower prices and avoided price rises. Still the entry ticket. In Deloitte's survey, improving margins is the top company-wide priority at 72%. Anyone telling you procurement is about to stop being asked about cost is wrong.

2. Money kept

This is cash rather than profit, and it is the one procurement most often forgets to claim. Payment terms. Early payment discounts. Stock sitting in a warehouse. The Hackett Group's 2025 working capital survey looked at the top 1,000 US listed non-financial companies and found $1.7 trillion tied up unnecessarily. That is 35% of their combined working capital. Average payment terms sat at 59 days.

Your CFO thinks about this every week. If procurement never mentions it, procurement is invisible in the conversation that finance cares most about.

3. Trouble avoided

Suppliers going bust. One supplier you cannot replace. A tariff you did not see coming. A contract that renewed itself.

QIMA surveyed more than 1,000 businesses that buy internationally. More than 90% of US- and China-based sourcing networks were hit by tariffs in 2025. Forty-three percent moved where they buy from. And most companies still cannot see far into their own supply chain: the average business has mapped 60% of its supplier network, and only 18% can see it end to end.

There is a growing reporting version of this too. CDP found that emissions from a company's supply chain average 26 times its own direct emissions. If your company has a climate target, most of it sits in decisions procurement makes.

4. Time saved

This is the one the rest of the business actually feels. How long from asking for something to being allowed to buy it. How long from invoice to payment.

Procurify looked at three years of data from more than 250 mid-sized companies. In 2025 it took about 58 hours to get from request to purchase order in companies of 100 to 500 people, and 65 hours in companies of 500 to 2,000. Their own guidance is that past 72 hours, the waiting itself becomes the problem.

Cut that from three days to one and every project manager in the building notices. They will call you earlier next time. Which brings us to the thing that matters more than all four.

#The single thing that decides everything: how early you get asked

Most of what a purchase will cost is decided before anyone talks about price. It is decided when someone chooses the specification, the material, the design, the delivery date. Widely used engineering estimates put around 70% of a product's total cost as locked in at the design stage.

So the time procurement is involved matters more than how hard procurement negotiates. One practitioner quoted by The Procurist put the difference at 20 to 30% higher quotes when procurement is brought in at the end rather than the beginning. That is one person's experience rather than a study, but it matches what the design-stage numbers imply.

Here is the uncomfortable part. You do not get invited earlier by writing a strategy document or asking for a seat at the table. You get invited earlier when going through you is faster than going around you.

Which means speed is not a nice-to-have that comes after strategy. Speed is how you earn the right to strategy.

#A twelve-month plan that does not need new software

Most published plans assume you are buying a system. That is why most of them stall around month four. This one assumes you have your existing finance system and nothing else, because that is where most mid-sized companies actually are.

Months 1 to 2: measure where you are

Twelve months of spend by supplier and category. How many invoices you handle and how many go wrong. How long a request takes to become a purchase order. How many suppliers you have, and how many you used more than twice. And an honest estimate of how your team's hours split between paperwork and real buying.

Do not start any initiative yet. A baseline taken after you have started is not a baseline. It is an argument.

Months 2 to 4: sort your spend into categories

Every pound or dollar assigned to a category your own team recognises. Anything you cannot classify goes into a bucket you report openly rather than hide. This one piece of work decides the quality of every analysis and every business case you make for the next three years. It is also the thing that decides whether AI will ever be useful to you.

Months 3 to 6: take the paperwork off your specialists

Invoices, purchase orders, checking requests, keeping supplier records clean, keeping catalogues current. You can automate this or hand it to someone else. Both are legitimate. What matters is that it stops eating the hours of the two or three people you need for category work, and that it starts being measured against agreed service levels instead of quietly absorbed.

Months 4 to 8: own the renewal calendar

Every contract, its end date, its notice period, and whether it renews itself. One list. One named owner. A review booked at least 90 days before each deadline.

This is the cheapest item on the list and usually the most profitable. Zylo's 2025 research found that 53% of software licences companies pay for sit unused, costing the average organisation around $21 million a year. Most of those renewed because nobody looked.

Months 6 to 12: spend the recovered hours on categories

Choose three categories by size and volatility, not by how easy they are. For each one, work out what the business really needs, what the market looks like, how the current supplier is performing, and what the alternatives are. This is the work only your people can do, and it is the work that has been squeezed out.

All year: report in four numbers, not one

Money saved. Money kept. Trouble avoided. Time saved. Agreed with finance in advance. One number invites one argument. Four numbers, agreed beforehand, turn the quarterly meeting into a review instead of a trial.

#Five things people get wrong about all this

"AI will make procurement strategic"

AI removes effort from tasks that are already well defined. It does not decide which categories matter. It does not tell a business unit its favourite supplier is failing. It does not hold a relationship together during a shortage. Where AI has really changed a procurement team, it did so by giving hours back. Those hours only become strategy if someone decided in advance what they were for.

"The savings era is over"

It is not. Margin is still the number one company priority in every recent survey of procurement leaders. A team that stops delivering savings in order to be strategic does not get promoted. It gets restructured. The change is additive. Savings plus three other things, not savings replaced.

"We need a platform first"

This is the most common way to lose a year. A source-to-pay system installed over sorted spend, clean supplier records and processes people actually follow is a good investment. The same system installed over none of those becomes a very expensive record of the mess. Comprara puts it well: a hospital does not design a treatment plan before running the diagnostic.

"We cannot find the right people"

Sometimes true. Often it is a design problem wearing a hiring problem's clothes, as The Procurist argues. The job description has quietly grown to include analytics, sustainability reporting, risk and AI, while the pay band, the tools and the title stayed where they were in 2015. Roland Berger's research on procurement in 2030 makes the same point from the other direction: nearly 90% of leaders say complexity has increased and 93% say stakeholder contact is at an all-time high.

"Outsourcing means losing control"

Only if the arrangement is built that way. Work done inside your own finance system, under your own approval limits, with your award decisions unchanged, leaves control exactly where it was. The test is simple. After go-live, does a purchase order still follow your rules and your sign-off, without anyone needing to explain an exception? Deloitte's own advice to procurement leaders includes using third-party managed services as one of the available options, with the ability to bring the work back in.

#What if you do not have a procurement team at all?

Many companies between 100 and 1,000 people do not. Buying sits inside finance by default, usually with someone who has another full-time job. Many organisations do not hire a dedicated procurement person until they pass a thousand employees. Most of the people who touch buying are department heads and office managers, not specialists.

If that is you, the advice above still applies, in a shorter form. You are not choosing between a good procurement function and a bad one. You are choosing between the work being done and the work not being done at all.

Start with three things. Know what you spend and with whom. Know when your contracts renew. Know how long it takes someone to get permission to buy something. Those three answers will tell you more than any strategy document.

#Where this leaves you

The large companies in these studies are answering the 2026 squeeze by spending more on technology. Deloitte's leading group now puts 24% of its budget into procurement technology, heading for 26%. A mid-sized company will not get that budget, and should not want it, because the licence is the cheapest part of that bill.

The same problem can be solved with a different tool. Buy capacity and clean data instead of licences, in the order set out above, so each step pays for the next one.

The procurement teams that look strategic in three years will not be the ones that bought the most software. They will be the ones that stopped spending most of their expert hours on work that never needed an expert, and then measured what those hours produced instead.

We have no client results to show you yet. That is exactly why the number that matters is your own, measured before anything changes.