Why Africa Still Wins the Outsourcing Argument

The machines came for the call centre. The call centre is doing fine.

In December 2024, Klarna's chief executive, Sebastian Siemiatkowski, told Bloomberg his company had stopped hiring because artificial intelligence "can already do all of the jobs". His AI assistant, he boasted, was doing the work of 700 customer service agents. By May 2025 he was recruiting humans again, conceding that the cost-cutting had gone too far and that "what you end up having is lower quality". Customers, it turned out, wanted to talk to people.

Klarna is the loudest example of a quieter pattern. Gartner, a research firm not given to sentimentality about labour, predicted in June 2025 that half of all organisations planning to cut customer service staff because of AI would abandon those plans. By September it went further: not one Fortune 500 company, it forecast, will have fully eliminated human customer service by 2028. And in the same season it estimated that over 40% of agentic AI projects would be cancelled by the end of 2027, victims of runaway costs and returns that failed to appear.

None of this means AI is a dud. It means the boardroom syllogism, that AI makes people redundant and therefore makes outsourcing redundant, has a faulty second step. The evidence points somewhere more interesting: AI is stripping out the drudgery, password resets, order lookups, the scripted tier-one queue, and leaving behind the work that is hardest to automate. Complaints that require judgement; retention calls that require empathy; and compliance checks that require someone accountable. That residue is smaller in volume but higher in stakes, and buying it well has become harder. Which is precisely the argument for buying it from Africa.

The numbers

Start with South Africa, the continent's most mature market. Its global business services sector employs roughly 150,000 people serving international clients, having added 20,518 net new offshore roles in 2024 alone. More than half of that workforce, some 55%, serves British clients, helped by a time zone one hour off London's and an accent British customers rate warmly. Enterprise buyers have noticed. In Ryan Strategic Advisory's annual survey of contact centre decision-makers, South Africa ranked first in the world for offshore customer experience in 2021, tied first with India in 2022, and remains the top preference among American buyers.

The economics remain blunt. A customer experience professional in Cape Town costs an employer roughly $14,000 to $18,000 a year; or $6,000 to $10,000 a year in Lagos, against $38,000 to $42,000 for the equivalent seat in Manchester, with all-in savings of 55% to 65% against UK, US, and Australian delivery. That is before government incentives, which add a further 7% to 10%. And the workforce stays put: annual attrition of 10% to 18% in Cape Town, or less than 4% in Lagos, against 30% to 40% in the Philippines and 30% to 35% in India. In a business where every departing agent takes their training budget with them, retention is margin.

The story is continental. Africa's BPO sector employs some 1.2 million people across more than 400 delivery centres, a figure projected to more than double by 2030. The capital is following the labour. CCI Global, which employs around 15,000 people across six African countries, opened a $50m campus at Tatu City outside Nairobi in 2024, East Africa's largest, with 70% of its business serving American clients. Amazon has run customer service from Cape Town for two decades and added 3,000 South African roles supporting customers in North America and Europe. These are not aid projects. They are procurement decisions made by companies with alternatives.

What AI changes

We have to honestly concede that there are a few things AI does destroy. Pure transcription, basic data entry, and the thinnest tier of scripted support are going, and providers still selling seat-hours against that work are selling a depreciating asset. The buyers' question has shifted from "how much per seat?" to "what does your blended human-and-AI operation deliver per outcome?"

That question favours Africa, for a demographic reason that no amount of model scaling alters. The continent has the world's youngest workforce, and the agent of 2030 is an AI operator, not a script-reader: someone who supervises automated workflows, handles the escalations the model cannot, and applies judgement where the stakes justify a salary. Training a 24-year-old graduate in Lagos or Durban to do that job costs a fraction of training one in Manchester, and Nigeria's pipeline alone produces around 400,000 university graduates a year, within the 1.7 million who leave its tertiary institutions annually. Its English is not merely serviceable. Nigerian candidates averaged band 6.7 on the academic IELTS in 2022, placing joint fifth among the top test-taking countries worldwide, and 7.2 in speaking, a score that beats nearly every country on the list. The regulatory scaffolding is already in place: South Africa's POPIA data protection regime is GDPR-aligned, and the country wrote the global contact centre standard, ISO 18295.

For a decision-maker, then, the question is whether, once the automatable 60% of the queue is automated, you want the remaining 40%, the judgement calls, the angry customers, the regulated conversations, handled in-house at Manchester prices, by a bot your customers will punish you for, or by a stable, educated, more affordable workforce in a time zone that overlaps your working day.

Klarna's epilogue is instructive. Its CEO now frames human customer service as a premium offering, a mark of quality the company advertises. The firms that automated their way out of human service are, one by one, paying to buy it back. Increasingly, the people selling it answer the phone in Cape Town, or in Lagos.

If you're looking to explore outsourcing to Africa and you're not sure where to begin, come talk to us at Ledgeris. We're happy to help, or to point you in the right direction.

---