For years, the question an operations director asked when outsourcing a process was always the same: how many people do I need? It was the natural unit of measurement. The provider quoted FTEs (full-time employees), the client approved FTEs, and success was measured in FTEs. But that model is reaching its limit, and more and more organizations are starting to ask a different question: what outcome do I need?
The FTE model has a structural problem
Traditional BPO is, in essence, a labor arbitrage business. The more work there is, the more people are hired. The more people are hired, the more the provider bills. The mechanics are simple, but they hide a contradiction that is rarely discussed openly.
The provider has no real incentive to become more efficient. If it automates a process, it loses revenue. If it drastically reduces errors, it needs fewer people to correct them. In the classic model, efficiency works against the provider's business. This is not bad faith; it is how the model is built.
For the client, this means a relationship in which inputs are bought (hours, people, capacity) when what is really needed are outputs: loans granted on time, claims resolved correctly, complaints closed without escalation.
From buying resources to buying outcomes
The change taking place in the most advanced outsourcing models is a deep one. Clients no longer want to buy FTEs. They want to buy outcomes.
And this is not just a change in pricing model. It is a change in responsibility. When an outcome is bought, the provider takes on the risk of efficiency. If it takes longer, the cost is theirs. If it makes mistakes, the cost is theirs. If volume scales, it has to absorb it without the client bearing more cost for each additional person.
This logic, which belongs to software, is now being applied to services. Hence the concept gaining traction in the industry: Services as Software. It does not mean people disappear from the process. It means the service behaves like software: with decreasing marginal cost, real scalability and clear performance metrics.
Automating tasks, not jobs
One of the most common mistakes when talking about automation is to treat it as a direct replacement of people. The most rigorous studies point to something more nuanced: AI automates tasks, not roles.
In a loan origination process, for example, there are routine tasks that a person performs repeatedly (data extraction, document validation, cross-checking of information) and tasks that require human judgment: analyzing exceptions, deciding on complex cases, communicating with the customer in sensitive situations.
Automating the former does not eliminate the role. It transforms it. The person stops doing what a machine can do faster and with fewer errors, and focuses on what adds real value. The result in real operations is significant: reductions of up to 80% in process times and savings of around 50% compared with a traditional BPO without technology. Not by eliminating teams, but by eliminating friction.
What to ask before choosing a provider
For an operations director or a CTO evaluating providers of outsourced services for critical processes, such as claims management, banking complaints or credit origination, there is a simple way to tell who has real technology and who is still selling people under a different label.
The first question is whether the provider charges per person or per outcome. If the answer is per person, the incentives are still misaligned. The second is whether it can scale without cost scaling proportionally. If every increase in volume requires hiring someone else, there is no technology underneath. The third is which SLA metrics it guarantees on the final outcome, not on intermediate activity. And the fourth, often forgotten, is whether the technology is its own or depends on third parties it does not control. When something fails in a critical process at three in the morning, the difference between proprietary technology and a stack of integrated third-party tools shows.
BPOs that cannot answer these questions convincingly will not disappear overnight. But they will lose ground to providers that combine people and technology to deliver the service with the economics of software.
Services as Software in practice
Services as Software is not a theory. It is a model that already works in critical banking and insurance processes: claims processing, customer onboarding, credit origination, complaint management. These are processes where volume is high, accuracy is non-negotiable and response times directly affect the customer experience.
At Serimag we have spent years building this model. Our own technology for document extraction and processing, specialized teams that bring judgment where technology falls short (humans in the loop), and a commitment to outcome metrics, not headcount. Because in the end, what matters is not how many people work on your process. What matters is that the process delivers guaranteed results.