AI has changed the economics of negotiation

A supplier asks for an 8% price increase, and procurement negotiates it down to 4%. Was that a good result?

You can't tell without knowing how the supplier's costs have actually moved and what refusing would cost you, given the alternatives you have. All the reduction records is how far the supplier came down from its opening demand.

AI is useful to procurement at exactly this point, because it can cut the effort of building a serious commercial position before the negotiation starts.

Preparation used to be expensive

A well-prepared buyer may need to analyse raw-material movements, labour and conversion costs, freight, historical prices, supplier performance, market benchmarks, previous concessions and alternative sources.

That effort is justified for a strategic category and often hard to justify for each of hundreds of smaller negotiations.

Digital tools and AI loosen that constraint. McKinsey reported a 2024 case in which Sanofi supported procurement decisions with should-cost modelling, analytics and digitally enabled negotiations. Lower spend, faster tender evaluation and higher negotiation savings were among the reported results.

AI does not produce savings like these automatically. What the case shows is that serious preparation can become economically viable across more negotiations.

Breaking down the supplier's request

This example is illustrative and does not describe a client.

A component currently costs €10.00. The supplier requests €10.80.

The buyer builds a cost-change bridge using documented assumptions.

On these assumptions the bridge implies a price of €10.34.

The figure is only as sound as those assumptions. What it gives the buyer is a better question to put to the supplier:

"Our current analysis explains €0.34 of the €0.80 requested increase. What explains the remaining €0.46?"

There may be a legitimate answer: capacity investment, compliance cost, a change in order economics, or an assumption the buyer missed.

The negotiation should bring that difference into the open and leave both possibilities standing until it is explained: the request may be justified, or the supplier may be wrong.

Analysis and bargaining power

Now suppose the supplier is the only qualified source and an alternative would take nine months to become operational. The buyer's analysis may be excellent and still give little bargaining power in the short term.

The buyer therefore has to keep three figures apart. Evidence gives the cost-based reference. The negotiation target is what the buyer means to achieve, and the approval limit belongs to the business: what it is prepared to accept once alternatives, continuity risk and the total commercial package are taken into account.

A buyer may rationally accept a temporary price above the cost-based reference in return for secured capacity or a defined review mechanism while developing another source.

AI can strengthen the buyer's argument, though the alternative supplier still has to be developed.

Use AI for the work it is good at

AI is suited to preparing and challenging. It retrieves authorised information, compares documents, identifies inconsistencies and proposes scenarios.

The calculations (cost movements, sensitivities, currency effects, offer comparisons) should run through controlled models with traceable formulas, and an accountable buyer approves the assumptions, concessions, limits and exceptions.

A count of the negotiations that used AI shows how much the tools are being used. The question for an executive is:

"Which commercial decisions improved, and what evidence supports that? What did we trade in return?"

Dov Amar · Opyflow