The End of Breakage

Agents will claim what customers never did, and the revenue that friction quietly produced is about to become visible

Margin that depends on inaction

A meaningful share of corporate revenue depends on customers not acting. They don’t submit the rebate, chase the SLA credit or challenge the invoice. They forget to cancel and accept the renewal. None of this shows up as a line in a forecast, but it sits inside the margin. AI agents do not leave these things undone.

This week’s piece, Agents Will Spend the Money in B2B, described how Access, Choice and Execution decide which suppliers an agent selects. An agent that buys on a customer’s behalf is equally well placed to enforce the contract afterwards. It holds the record of the price, the service levels, the cancellation terms and the credits owed, and it has no reason to let any of them lapse.

Friction has been a revenue line

The industry term for this is breakage. Common examples are the price-match difference judged not worth twenty minutes on hold, the service credit nobody claimed for a missed SLA, and unused seats that renewed because nobody checked the usage. Each one is small. In aggregate they form a source of profit that most companies have never measured, because measuring it would require asking how much revenue comes from customers failing to exercise their rights.

Breakage depends on a person bearing the cost of effort, and an agent removes that cost. It reads the full policy every time, finds the relevant clause and applies it. It remembers what was promised in earlier conversations. It will wait on hold, escalate and try again the following day. Claims that a person would have dropped become routine, and they arrive together: credits, invoice disputes, switching, cancellation and every other contractual right a customer holds, all exercised by the same tireless system.

The immediate effects fall on customer service and finance. Refunds and concessions rise. Contact volume and cost to serve rise with them, at a time when many service teams are being asked to shrink. Outsourcing contracts priced per contact, and capacity plans built on historical abandonment rates, assume a customer population that is about to behave differently. Service teams also spend more of their time negotiating with software, and lose part of the relationship they exist to protect.

In B2B, the loop runs in reverse

The earlier piece described a compounding loop. A supplier that passes Access, wins Choice and completes Execution becomes part of the organisation’s operating pattern and tends to be reused. Part of what made that loop durable was inertia. Contracts renewed because nobody reviewed them, and credits went unclaimed because finding them took an afternoon.

A procurement agent that reconciles invoices against contract terms removes that inertia. It can flag every missed service level, claim every credit, compare renewal pricing against alternatives it already knows and recommend cancelling seats with no usage. The machine reader that put a supplier on the shortlist is also auditing whether the supplier delivered. For SaaS vendors this arrives at renewal, where net revenue retention is decided.

The contract becomes executable

Humans interpret contracts. Agents execute them. When the first reader of a refund policy, service agreement or cancellation clause is a machine, those documents behave like code that will run exactly as written, every time it is triggered. Any gap between what the policy says, what the sales team promised and what support routinely concedes becomes an inconsistency the agent will find and use.

The same terms matter before the sale. Agents evaluating suppliers can read them during Choice, so a clear and competitive policy may help a vendor win selection, while an unclear one can cost it twice: once at selection and again at claim.

Contract design will change

The longer-term consequence is larger than rising refunds. Today’s policies were priced on the assumption that only a fraction of customers would exercise their rights. If agents exercise those rights routinely, the expected cost of every guarantee, credit and refund policy changes. Vendors will respond by tightening terms, simplifying SLAs, raising prices or withdrawing guarantees until the expected claim rate is affordable again.

This changes the basis of competition. Companies have long competed on the promises they make. Increasingly they will compete on promises they can afford to have enforced by software, every time, for every customer. Generous terms that worked because few people used them will become expensive, and vendors able to offer strong guarantees at full uptake will hold a real advantage.

What CX leaders and CFOs should measure now

The three gates apply after the sale as they do before it.

At Access, can agents reach the service and cancellation routes, and through which channels? A vendor that blocks agents entirely may find the dispute moves to a chargeback, a human escalation or a competitor.

At Choice, when an agent reads the vendor’s terms during evaluation, how do they compare with the alternatives, and do they help or hurt selection?

At Execution, when an agent pursues a claim, what does it obtain, how long does it take, and how far does the outcome depend on persistence rather than on the policy itself?

Finance teams should add a fourth question: how much current margin rests on claims customers do not make? That figure is the exposure.

Every CFO knows how much revenue comes from selling. Few know how much comes from customers not exercising their contractual rights. Agents will expose that number, and some of what now appears as margin may turn out to have been friction.