When Your Supplier Automates the Work, Who Gets the Savings?

Imagine renewing a business-process outsourcing agreement that has been priced for years around people: a monthly charge for a team, an hourly rate for additional work and an annual uplift to cover wages. The supplier has since introduced software that classifies requests, prepares replies and closes a growing share of routine cases without human intervention. Service levels look acceptable, but the invoice is still organized as if the old delivery model were intact. At renewal, the supplier asks for another increase.

The buyer's first instinct may be to demand that the supplier pass on its AI savings. That sounds reasonable, but it skips the question the negotiation actually needs to resolve: which savings exist, who created them, who paid for the investment and what does the contract say the customer has purchased?

An HFS Research operational playbook published on 12 October 2026 addresses that challenge for retail and consumer packaged goods outsourcing. Its central proposal is to update the scope and pricing model as automation changes the work, rather than relying on another discount against an outdated rate card. The framework is specific to BPO arrangements, but the underlying commercial problem can appear in managed services, customer operations and other contracts built around effort instead of delivered outcomes.

A cheaper delivery process does not automatically mean a cheaper contract

If the supplier committed to delivering an agreed service for a fixed fee, then investment in productivity may have been part of its commercial risk. The customer cannot assume every internal efficiency improvement belongs to them. Conversely, if the price explicitly reimburses named full-time equivalents or billable hours, continuing to pay for effort that is no longer required raises an entirely different contractual question.

A buyer therefore needs to distinguish the contractual pricing unit from the operational resources used to deliver it. Suppose a service once required twenty people to handle ten thousand cases each month. After automation, the supplier uses twelve people, but the demand mix, exception rate and quality expectations may also have changed. The raw headcount difference tells only part of the story. The team might now be resolving more difficult cases, supporting longer opening hours or carrying new compliance work. Before discussing an entitlement to savings, procurement must establish which of these changes are real.

This is why a percentage discount is often too blunt. If both parties are still arguing about the price of the old unit of effort, they may never reach the more important question of what unit should be priced next.

Build the baseline before arguing about productivity

A defensible renewal starts with a shared picture of the service. Procurement and the operating function should compare invoices, volumes, rate cards, agreed scope and service results over a meaningful period, ideally including the time before and after automation was introduced. They should understand what share of the work follows an automated route, which exceptions still require specialists and whether the customer now receives something materially different.

HFS recommends establishing unit economics by process and quarter, checking billed resources against available operational evidence and identifying whether apparent productivity is being obscured by changes in demand or case complexity. A sensible commercial discussion might therefore focus on the cost per completed case, correctly handled return or verified transaction rather than the number of people assigned to an account.

The baseline matters because both sides can otherwise make plausible but incomparable claims. The provider points to rising complexity and investment costs; the customer points to fewer human touches. Without agreed units, neither claim can be tested.

Negotiate the investment and the future savings together

Automation can generate gains for both parties, but the distribution needs to be designed. If the supplier paid to develop and deploy the automation, it may reasonably seek a period to recover that investment. If the customer funded the system, contributed proprietary workflows or guaranteed the volumes required for deployment, the buyer has a stronger argument for a larger share of the benefits. Neither position should be accepted as a slogan; both require evidence.

One practical approach is a defined recovery period, followed by an agreed productivity adjustment to the relevant pricing units. Another is a gain-sharing mechanism that compares verified unit cost or throughput against a baseline and allocates a specified share of the improvement to each party. Where measurable outcomes exist, the contract can gradually shift a suitable part of the service from effort-based to transaction, consumption or outcome pricing.

Those mechanisms need boundaries. Who owns the measurement data? How are quality failures and rework counted? What happens when the volume falls, the case mix changes or regulation introduces additional manual checks? A gain-sharing formula that ignores these effects will eventually become another dispute.

The supplier's objections deserve commercial answers

A supplier may argue that automation is its intellectual property and that the economic benefit should remain with the party that financed it. That is a legitimate opening position, not the end of the negotiation. The buyer can ask which investments are account-specific, which were already recovered elsewhere and whether both parties would benefit from a transparent transition to a new pricing unit.

The supplier may also claim that its automation details are commercially sensitive. Procurement seldom needs access to the source code or internal model weights. It needs evidence about the proportion of work completed without human handling, the consistency of that measure and the services for which it is relevant. Disclosure can be bounded by process, audited under confidentiality and separated from proprietary implementation details.

A third objection concerns outcome pricing. Suppliers may charge a risk premium when outcomes depend on customer inputs or when volumes and quality are poorly measured. Rather than forcing an immediate conversion, the parties can instrument a narrow process, run a trial pricing unit and agree in advance what evidence will trigger the wider transition. A credible alternative or a small competitive carve-out may also establish an external price reference without threatening the entire service overnight.

The negotiation works better when the buyer treats these objections as hypotheses to test and possible trades to structure, rather than as proof that the supplier is acting in bad faith.

Write future automation into the agreement

The most valuable clause may not be the first price reduction. It may be a mechanism that stops the same debate recurring whenever new automation is introduced. A mature agreement can define how process automation is disclosed, how units and service levels are measured, when pricing is reviewed and how improvements are reflected in future charges. Data access, auditability, exit rights and ownership of customer-specific knowledge bases or configurations matter because they determine whether the buyer can verify performance and switch providers if needed.

Not every service can be priced neatly by outcome. Some require standby capacity, specialist expertise or flexibility during peaks. The point is to match the commercial model to the work that actually takes place, preserving valuable flexibility rather than erasing it through a simplistic pursuit of the lowest unit cost.

For procurement, the difficult part is not knowing that automation can lower cost. It is holding a commercially coherent conversation about evidence, investment, risk and reciprocal value when the provider rejects the first proposal. That is exactly the kind of exchange buyers should rehearse before the renewal: the supplier has plausible reasons to protect its margin, while the customer has equally plausible reasons not to fund yesterday's delivery model indefinitely.

AI does not make the productivity gain automatically belong to either party. It changes what can be negotiated. The buyer's task is to recognize that change before an obsolete pricing model becomes the next multi-year commitment.

Sources

• HFS Research (12 October 2026). Operational Playbook: RCPG must re-commercialize BPO contracts around outcomes and automation disclosure.
• Voice2Evolve Learn: Why Procurement Savings Do Not Reach the P&L.