Skip to content
Exclusive partner BME Symposium18–19 November · Berlin · Booth A16Book a meeting
Contact

Procurement value is decided before the signature.

Order processing determines what a purchase order costs. Strategy, sourcing and negotiation determine what the spend costs. The second lever is an order of magnitude larger.

In short

Strategic procurement covers two phases from demand to signed contract. Plan-to-strategy covers demand analysis and category strategy; source-to-contract covers market and supplier analysis, tendering, negotiation and the contract. This is where it is decided what is bought, from whom, at what price and with what risk. Purchase-to-pay lowers the cost of processing. The path from strategy to contract acts on the entire spend and therefore has the larger lever on profit.

Data: third-party research with source. Our view: the joint position of H&Z and Elevatry, indicative observations from project work. Model: calculations and examples, not customer results.

Two levers, two orders of magnitude.

Procurement value can be pursued from two sides. The transaction lever lowers the cost per transaction: ordering, goods receipt, invoice checking. The volume lever changes what is bought, from whom, at what price, with which index clause and which notice period. Both are legitimate. They simply act on bases of very different size.

The transaction lever is bounded by the cost of processing. Even fully automated order processing cannot save more than it costs today, usually a fraction of one percent of spend. Even if the entire procurement process disappeared, our model with €500M of spend would save €3.5M a year. The volume lever, by contrast, scales with spend: 5 to 10% better outcomes are worth €25M to €50M, many times that ceiling. McKinsey puts the difference at roughly elevenfold.

That implies a sequence. Transactional automation pays off where a strategic decision has already created value, because it protects that value. Starting with processing because it is easier to automate means starting with the smaller lever. Processing still matters, but it follows the decision, not the other way round.

Figure 1ModelThe transaction lever stops at process cost. The volume lever grows with spend.

Efficiency ceiling

€2.1M
€3.5M
€25M
€50M
Transaction: processing fully automatedTransaction: procurement process eliminatedVolume: 5% better outcome on the spendVolume: 10% better outcome on the spend

Model, the same sample company as on the AI in procurement page. Assumptions: €500M addressable spend; procurement process cost 0.7% of spend (€3.5M), of which 60% is operational processing (€2.1M). Volume lever 5–10% of spend per the H&Z and Elevatry lever model. Values per year, before implementation cost.

Every phase ends with a decision that is not delegated.

The path from strategy to contract has six stages – spend & demand, category strategy, market & suppliers, tendering, negotiation and contract – and each closes with a human decision. The team confirms the baseline, chooses the levers, releases the supplier list, approves the shortlist, leads the negotiation, awards and signs. In between lies the depth: profiling spend, breaking down cost drivers, drafting tender documents, turning bids into scenarios, preparing the BATNA and the concession ladder.

That depth is what Companions take on. How far they may go depends on reversibility and materiality. Deadline monitoring or spend classification is cheap to correct, so the Companion runs in autonomous mode with approval. An award or a strategic change of supplier is hard to undo, so it prepares and a person decides.

Leading the negotiation and making the award stay with people by design. This is where accountability for commitments, the quality of supplier relationships and often the trust of the works council sit. A well-prepared negotiator with a clear fallback position is stronger than one who had to do the groundwork alone.

74%of CPOs' time goes into transactional and operational work, and still 63% at top performers.DataDeloitte Global CPO Survey 2025
Companions carry the depth. People carry the decisions.

Value leaks between the phases, not within them.

Identified potential is not yet a result. It has to be approved, negotiated, written into the contract, entered into master data and held over the term. Something is lost at every stage: measures are not pursued, concessions are given up in negotiation, agreed prices never reach the system, index clauses are never invoked in the buyer's favour.

The cause is usually a broken thread. The reasoning behind the strategy never reaches the negotiation; the negotiated result never reaches master data. Two years later the same project is tendered again, because nobody knows what was decided last time and why.

Building each phase on the result of the previous one, and keeping decisions with their reasons, closes these gaps. What gets measured is then realised value against a versioned baseline, not hours saved and not identified potential. Only that number shows whether procurement actually moves the result.

Figure 2ModelOf €8.0M in identified potential, €3.1M reaches the bottom line.
Identified−2.1€8.0M before the next stage€8.0M
Approved−1.5€5.9M before the next stage€5.9M
Contracted−0.8€4.4M before the next stage€4.4M
Implemented−0.5€3.6M before the next stage€3.6M
Realised€3.1M

Model of one year of a strategic sourcing programme, not customer data. Conversion rates are assumptions and vary widely; what matters is measuring each step.

Companions for this topic

What Elevatry does for you.

Companions work around the clock in your data, with the H&Z method from more than 5,000 procurement projects. They prepare decisions, work in your experts' feedback and, once you approve, carry out the next steps. Your team decides.

All Companions
Model

A tender for machined parts, phase by phase

An engineering company buys turned and milled parts worth €14M a year for two plants from nine suppliers. The framework agreement with the largest supplier ends in ten months. A new product line will raise demand by about 18%.

  1. 01
    Demand and baseline

    The Companion profiles 140 part numbers, adds the new product line and proposes a baseline per part. The team confirms scope and baseline.

  2. 02
    Strategy and levers

    Machine hours account for about 45% of cost, material for about 35%. There is spare capacity in the region. The proposal: bundle across both plants and request quotes with an open cost breakdown.

  3. 03
    Bids as scenarios

    Six suppliers bid. A single award would land 5.3% below baseline, a dual-sourcing option 4.5% below, with markedly lower supply risk.

  4. 04
    Negotiation and contract

    The steel price clause is tied to the material share only, not to the whole part price. Notice periods and price adjustments go into the deadline calendar.

  5. 05
    The team awards

    Single award or two sources is the team's call, based on its appetite for risk. The Companion has put a number on both.

With dual sourcing, about €0.63M a year against the baseline, plus an index clause that now acts only on the actual material share.

Illustrative, not a customer result.

Questions worth asking.

  1. 01

    How much of our procurement time goes into the phases before the contract, and how much into processing?

  2. 02

    What share of identified savings actually reaches our P&L?

  3. 03

    In which phase do we lose the most knowledge when a category manager leaves?

  4. 04

    Who decided which steps before the contract a machine may prepare, and which stay closed to it?

Free download

Practical guides on this topic

The Elevatry guides take the topic into day-to-day work: as a PDF, free of charge.

What procurement leaders ask us.

Should we use AI in order processing or in strategic sourcing first?
In strategic sourcing, because processing is the smaller lever and it has a hard ceiling. Automation can save at most what the process costs: in our model with €500M of spend, at most €3.5M a year, even if the entire procurement process disappeared. 5 to 10% better outcomes on the volume are worth €25M to €50M in the same model. So the sequence is clear: first improve the decision on supplier, price and index clause, then automate the processing that protects that value.
Why do identified savings not show up in the P&L?
Because savings leak at the handovers, not in the negotiation. The agreed price never reaches ERP master data, requisitioners buy around the new contract (maverick buying), index reductions are never claimed, and volume is not steered to the tender winner. In our model, only €3.1M of €8.0M identified potential reaches the P&L. That is a memory problem. Building each phase on the previous one and measuring realised value against a versioned baseline closes the gap.
How do we introduce AI in procurement without launching an IT project?
Start with a decision, not a rollout. A framework agreement expiring in ten months is a better starting point than any data clean-up: it has a deadline, a euro figure and a named decision-maker. The data that case needs is improved in parallel, not up front. One well-made award convinces a board more than three pilots in order processing. After that, the same method is applied category by category, without the team having to grow.
Which steps from strategy to award can AI take on?
Everything between the decisions, none of the decisions themselves. Companions profile spend, break down cost drivers, draft tender documents, turn bids into scenarios and prepare the BATNA and concession ladder. Confirming the baseline, choosing levers, approving the shortlist, negotiating and awarding stay with the team. The line is drawn not by technology but by reversibility: a classification is cheap to correct, an award binds for years. What is new is that this depth is no longer reserved for large tenders.
How much can AI save in a sourcing event?
In our worked example, about 4.5% below baseline, and that value comes from better levers, not faster documents. On €14M of machined parts, that is roughly €0.63M a year. It comes from bundling across two plants, open cost breakdowns and a steel clause that now applies only to the material share. A single award would reach 5.3%, with higher supply risk. What is new is that both options carry a number before the team awards.

Friday notes on the topic

Next step

See it on your own procurement.

We show you on one of your processes how Companions do the work from analysis to contract. With your data and your approval rules, if you like.

See the platform