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

The 80/20 rule was a law of capacity.

Procurement concentrates its experts on a fraction of suppliers because expertise is scarce, not because the rest is worthless. A spend analysis that reads the whole portfolio moves that line.

In short

Spend analysis organises all of a company's external spend by category, supplier, plant and contract and shows where savings potential and risk sit. Traditionally it focuses on the largest items, because there is no capacity for more. With AI, the entire portfolio can be analysed regularly. What then matters is not savings alone but the share of spend under active management.

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.

Pareto allocates attention, not value.

ABC analyses, portfolio matrices and Pareto cuts pass for analytical tools. In practice they are rules for allocating scarce expert time. The largest categories get cost models, multi-round tenders and carefully prepared negotiations. Everything else is renewed, re-ordered and re-approved.

That was rational while every analysis cost days. But it has a side effect: the rule confirms itself. Where nobody looks, no potential is found, so the rest is deemed unproductive, so nobody keeps looking. The 80/20 split then describes not where the value is, but where the time goes.

When the price of analysis changes, so does the calculation. Once Companions take on much of the knowledge-intensive analysis, reviewing the whole portfolio becomes affordable. The question is no longer which suppliers you can afford to look at, but which ones you can afford to overlook.

71%of spend is actively managed on average; the rest manages itself.DataArdent Partners, The Metrics that Matter in 2025

Unmanaged value hides in plain sight.

The value outside the top suppliers rarely sits in one big find. It sits in many small patterns: six prices for one specification from six suppliers. Index clauses never invoked in the buyer's favour when raw material prices fall. Auto-renewals because nobody worked the notice period backwards. Price differences between plants that persist year after year.

Each case on its own is minor. Together they form a structural leak that no savings programme reaches, because none had the capacity to look. A spend analysis that only sorts volumes will not find these patterns. For that, invoice lines, specifications and contract clauses have to be read together.

Figure 1ModelFour small leaks in neglected spend add up to a seven-figure sum.
€0.9M
€0.8M
€0.9M
€1.0M
€3.6M
Price spread for the same specificationIndex clauses not invokedAuto-renewals without renegotiationPrice differences between plantsTotal per year

Model. Assumptions: €500M spend, of which €200M is not actively managed. Price spread 1.5% on €60M; index decline of 4% on €40M, half not passed on; 3% on €30M of auto-renewed contracts; plant gap 2% on €50M.

What gets reported as a savings gap is often a coverage gap.

Measure coverage, not only savings.

Savings and cost per purchase order say little about how much of the portfolio is actually shaped. The fitting measure is managed-spend penetration: the share of external spend that receives expert-grade attention at each decision point. Spend counts as actively managed when it has a current strategy with defined levers and has been part of a structured sourcing or negotiation event in the last 24 months. According to Ardent Partners, 71% of spend is under active management on average; every additional euro brought under management saves 6 to 12% in the first contract term.

The measure makes the coverage gap visible and changes the conversation in the boardroom. What gets measured gets staffed, by people or by machines. Next to it belongs realised value per euro of spend, so that coverage is not confused with activity.

More coverage on its own is not enough. AI without procurement method produces more analyses with average reasoning, and that is risky in a negotiation. Only when the method the top categories get today is applied to the rest of the portfolio do coverage and quality rise together.

60–70%of knowledge-intensive, complex analytical work can be taken on by Companions.ModelTarget value from Elevatry analysis and H&Z project work, not a customer result

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

Six prices for one standard part

A manufacturer buys pneumatic fittings and standard parts worth €2.4M a year for four plants from 23 suppliers. The category has no owner. Orders go through free text and old framework agreements.

  1. 01
    Classify

    The Companion maps about 9,800 invoice lines to 310 specifications. A third of the lines had no material number.

  2. 02
    Compare prices

    The same specification is bought at up to six different prices. The median spread between cheapest and most expensive is 18%.

  3. 03
    Read the contracts

    Two framework agreements renew automatically. For one of them, the notice period ends in ten weeks.

  4. 04
    Value it realistically

    The benchmark is the second-cheapest price per specification, not the cheapest. That yields about 11% of volume, without assuming any change of supplier.

  5. 05
    The category manager decides

    Whether to bundle, steer through a catalogue or first just renegotiate is a human decision, made with plants and supply in mind.

About €0.26M a year in a category nobody had worked on, and an auto-renewal that can be stopped in time.

Illustrative, not a customer result.

Questions worth asking.

  1. 01

    What share of our spend has a current strategy with defined levers, and what share is simply renewed?

  2. 02

    How many different prices do we pay today for our ten most frequent specifications?

  3. 03

    Which index clauses would have worked in our favour in the last twelve months, and who invoked them?

  4. 04

    Do we report savings or coverage to the board, and what would change if we reported both?

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.

What data does AI need for a spend analysis?
Less than most assume: invoice or purchase order lines, plus contracts and, where available, specifications. What matters is reading them together. Only then do six prices for one specification show up, along with an index clause never invoked when raw material prices fell and a renewal whose notice period ends in ten weeks. Missing material numbers are no obstacle; in our worked example, a third of 9,800 lines had none. The whole portfolio becomes readable, not just the clean part.
How much can an AI spend analysis realistically save?
In our model, about €3.6M a year on €500M of spend, and it sits where nobody looks today. The amount does not come from the A categories but from the €200M without active management. It is made up of four small leaks: price variance for the same specification, index clauses not invoked, automatic renewals and price gaps between plants. So your figure depends less on the AI than on how large your unmanaged share is.
Do we need a master data project before an AI spend analysis?
No, and starting there pushes the first euro out by years. A master data programme without a concrete lever has no stopping criterion and no sponsor measured on results. The better route is the reverse: start with a quantified lever, such as price variance on standard parts, and improve exactly the data it needs. In our worked example, the Companion maps 9,800 invoice lines to 310 specifications. Data quality then grows with every decision, not ahead of it.
Which KPI shows the board how much spend procurement actually manages?
Managed spend penetration: the share of external spend with a current strategy, defined levers and a structured tender or negotiation in the last 24 months. According to Ardent Partners, this share averages 71%. Savings reports measure only the categories that were worked on. They are silent about the rest, and in our model with €500M of spend, that is where a €3.6M annual leak sits. What gets reported as a savings gap is often a coverage gap. What gets measured gets staffed, by people or by Companions.
Is spend analysis worth it for C-parts and small suppliers?
Yes, precisely there, because nobody has looked so far. The 80/20 rule was a limit of capacity, not of value. In our worked example, a manufacturer buys standard parts worth €2.4M from 23 suppliers, with up to six prices for the same specification and a median spread of 18%. Moving to the second-cheapest price per specification yields about €0.26M a year, without changing suppliers. Once Companions take on the knowledge-intensive analytical work, it pays off for every category.

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