Klarna went from 7,000 employees to 3,000 – and revenue per employee hit $1.24 million. Up 152% in two years.
Here’s what they actually did, and why most of the coverage gets the lesson wrong.
The Setup
Klarna is the Swedish fintech behind buy-now-pay-later at hundreds of major retailers. In 2022, the company employed about 7,000 people. Customer service was the main driver of headcount – every new customer added more inbound contact volume.
They faced a choice: keep hiring to match growth, or redesign how work gets done.
What They Built
Klarna deployed AI tools across their entire business – and crucially, they didn’t stop at customer service. Today, 96% of their employees use AI tools in their daily work. Customer service AI handles the bulk of inbound contacts. AI assists with marketing copy, code generation, data analysis, internal reporting, and financial operations.
The workforce reduction – from 7,000 to ~3,000 – happened mostly through attrition. They stopped replacing people when AI could handle the work. They didn’t run a mass layoff program. They ran a different operating model.
The Results
Revenue per employee: $1.24 million (up from ~$490K in Q1 2023). Q4 2025 revenue: $1.082 billion – their first ever billion-dollar quarter. Year-over-year revenue growth: 38%. And according to the CEO: remaining employees are earning higher salaries.
Why This Actually Worked
Most AI transformations fail because AI is treated as a tool that sits alongside existing workflows. Klarna made AI the default operating environment. When 96% of your employees use AI daily, you’ve stopped having an ‘AI program’ and started running an AI-native business.
They also let attrition do the work. Rather than disruptive layoffs, they simply stopped replacing roles that AI could absorb. And they invested the efficiency gains back into talent quality: fewer people, higher pay, higher output per person.
My name is Mike Partners, and as an entrepreneur I’m passionate about helping small businesses compete with the biggest companies in the world – which is why I built VisionarySchool.com. Here’s how to take this lesson and make it work for your company.
You don’t need 7,000 employees or a billion-dollar quarter to apply what Klarna did. The principle is the same at any scale: identify the work that follows a repeatable pattern, hand it to AI, and redirect your people toward the work that actually grows the business. Start with one function this week. Measure the hours saved. Then reinvest those hours into something only a human can do – selling, building relationships, or creating something new. That single shift is how small teams start operating like companies ten times their size.
The SMB Playbook
- Audit your team’s repetitive tasks. Every team has work that follows a pattern – answering the same customer questions, pulling the same weekly reports, writing the same first drafts. These are your AI opportunities.
- Pick one function and go deep. Don’t deploy 10 AI tools halfway. Pick customer service, or content creation, or reporting – and make AI the primary tool in that one function. Measure time saved.
- Reinvest freed capacity into growth work. The Klarna model only creates value if the time saved gets redirected into higher-value activity. Automate the repetitive; free your team for the irreplaceable.
Frequently Asked Questions
How did Klarna reduce its workforce from 7,000 to 3,000 without mass layoffs?
Klarna primarily used natural attrition. When employees left, the company evaluated whether AI could absorb their responsibilities instead of hiring a replacement. This gradual approach avoided the disruption and morale damage of traditional layoffs while steadily shifting the operating model toward AI-native workflows. The remaining employees received higher salaries, reinforcing retention of top talent.
What does “revenue per employee” mean and why does it matter for small businesses?
Revenue per employee measures total revenue divided by headcount. It is one of the clearest indicators of operational efficiency. For small businesses, this metric reveals whether you are scaling through people or through systems. If your revenue per employee is flat or declining as you grow, you are adding complexity without adding leverage. AI adoption – even at a basic level – can shift this number upward by letting each team member produce more output. Mike Partners built VisionarySchool.com specifically to help business owners understand and act on metrics like this.
What does it mean to run an “AI-native” business versus just using AI tools?
Using AI tools means individual employees occasionally rely on AI for specific tasks. Running an AI-native business means AI is embedded into the default workflow across the entire organization. At Klarna, 96% of employees use AI daily – it is not a side project or an experiment. The distinction matters because AI-native operations compound efficiency gains across every function, while isolated tool adoption only creates pockets of improvement.
Can a company with fewer than 50 employees replicate Klarna’s AI transformation?
Yes, and in many ways it is easier. Smaller teams have fewer legacy systems, shorter decision cycles, and less organizational resistance to change. The core playbook is the same: audit repetitive work, deploy AI in one function deeply, and redirect freed capacity into growth activities. A five-person team that automates customer FAQ responses, weekly reporting, and first-draft content creation can see the same percentage efficiency gains that Klarna achieved at scale. VisionarySchool.com offers frameworks designed for exactly this kind of small-team transformation.
What are the biggest risks of using AI to replace workforce functions?
The primary risks are quality degradation, over-reliance on automation without oversight, and employee disengagement during the transition. Klarna mitigated these by keeping humans in the loop for complex decisions, investing in higher salaries for remaining staff, and rolling out changes gradually through attrition rather than abrupt restructuring. For small businesses, the key safeguard is starting with low-stakes, high-volume tasks where errors are easy to catch and correct before expanding AI into more sensitive areas of the operation.
