Just how much can artificial intelligence transform a fintech’s core marketing and customer service functions — and its bottom line? Klarna, a global BNPL online shopping company, offers a real-time, rapidly evolving case study.
The Stockholm-based e-commerce company, which filed for a U.S. IPO on November 13, has been an aggressive adopter of the disruptive technology ever since OpenAI launched its first generative artificial intelligence chatbot, ChatGPT, in November 2022. Klarna became the first global fintech, and the first company in Europe, to use a Chat GPT “plug-in” to integrate OpenAI’s large language model into its own systems, giving shoppers personalized recommendations on everything from sneakers to mobile phones through a conversational interface focused on a consumer’s budget and tastes.
This year, Klarna has doubled down on AI once again.
The company says that its deployment of AI software in the first quarter of 2024 allowed it to run more ad campaigns with more images, all at a lower cost. It used generative AI to create more than 1,000 images during the first three months of this year, and slashed the time needed to produce images, including checks for brand consistency, image quality and legal compliance, to seven days from six weeks. It also shaved $6 million off image production costs by using AI tools from San Francisco-based Midjourney, OpenAI’s DALL-E platform and Firefly, part of Adobe, along with Topaz Gigapixel and Photoroom for final adjustments.
With 85 million active shoppers across 26 countries, Klarna is a bellwether for financial services and consumer goods companies, from competitors including PayPal, AfterPay and Affirm to, potentially, multinational giants like Procter & Gamble. In the first half of this year, shoppers spent an annualized 1 trillion Swedish Krona ($924 billion) on goods and services sold by nearly 600,000 retailers through Klarna.
“Klarna’s sophisticated use of customer data puts it at the leading edge compared with traditional banks,” S&P Global wrote in April, citing the company’s AI-driven insights into customers’ payment behavior that allow it to constantly tweak its credit scoring models.
The loss-making company, which is expanding rapidly in the United States and branching out into traditional financial services such as savings accounts, credits AI with helping to steer it toward profitability. In the second quarter of this year, Klarna posted a loss of 10 million Swedish Krona (nearly $923,000) compared to a loss of 854 million Swedish Krona (almost $79 million) in the year-ago period. The fintech, whose backers include Softbank, Sequoia and Heartland, was valued at around $14.6 billion as of late October.
AI has allowed the company to reduce its spending on sales and marketing by 11% in the first quarter of 2024. Klarna’s savings came to the detriment of third-party marketing agencies, on which the company slashed its spending by 25%. Overall, AI fueled 37%, equivalent to $10 million a year, of the company’s total cost savings for the quarter.
Last January, Klarna rolled out an AI “agent” — a souped-up chatbot — powered by OpenAI. The technology handled two-thirds of Klarna’s customer service queries about returns, refund, cancellations and payments, or 2.3 million conversations, in its first month of its operation, slashing the average time needed to resolve a customer query to two minutes from seven minutes and reducing the number of repeat inquiries by 25%. The AI agent, which functions in 35 languages, will boost Klarna’s profits this year by $40 million USD.
AI agents — like Klarna’s customer service chatbot embody — so-called autonomous intelligence, the most developed iteration of Gen AI after assisted, augmented and automated intelligence, all earlier AI versions that are used to screen resumes or power Alexa, Google Home and banks’ customer service chatbots. Such agents can also be used for credit analysis, investment recommendations, enhanced risk management, project management, and front-end and back-end software development. Because they can understand and process nuanced queries and learn from interactions with customers and users, they provide tailored responses that evolve over time to reflect the iterative refinements. Moody’s calls AI agents a “game changer” for the financial services industry that “could potentially create a new standard for excellence in financial analytics.”
Klarna offers installment payment plans to consumers buying goods and services from companies including Uber, Nike, AirBnB and Macy’s, which pay Klarna a 3%-6% commission for driving shoppers to buy on their websites. It has used AI to weed out or reduce offerings from more than 2,500 retailers that sell on its website.
But the company’s use of AI goes far beyond marketing and customer service. It also extends deep into the company’s structure and operations.
In August, CEO Sebastian Siemiatkowski told a second-quarter earnings call that Klarna was jettisoning its use of Salesforce, Workday and other widely used software-as-a-service (SaaS) providers of customer relationship management, human resources and payroll services. Klarna plans to use AI to build its own automation technology in-house more cheaply.
The news sent shockwaves through the industry, with Salesforce CEO Marc Benioff questioning in September how Klarna would build, achieve and maintain compliance and governance structures with in-house software.
Siemiatkowski is brushing off those risks. “I don’t think this is the right time to go all in and put all of your eggs into one of these platform baskets,” he told Digiday in May.
From 86% to 93% of the Klarna’s communications, marketing and legal teams use AI in their everyday work. Each of the company’s more than 1,900 software engineers and front-end developers uses an AI “co-pilot” assistant to generate and review the computer code embedded in the company’s website. The average revenue per employee as of June 30 of this year shot up 73% from year-ago levels.
Klarna’s use of AI for its advertising campaigns threatens to upend a decades-old industry of creating prototypes, storyboards and test shoots, and hoping the end results come out as wished. “Three weeks in and $100,000 later, I might see it and not like the idea,” Siemiatkowski told Digiday. Now what we do is sketch it within an hour,” he said, adding, “Ask Procter & Gamble or Pepsi how much they’ve spent on storyboards.”
On at least one level, the gains in productivity and efficiency have a cost. Klarna says its customer-service AI agent is doing the work of 700 full-time humans. The company, which has trimmed its workforce to 3,800 from 5,000 staff over the past year, aims to get its headcount down to 2,000 workers, Siemiatkowski told The Financial Times in August. The jobs cuts and cost reductions from AI, he added, had boosted average annual revenue per employee to around $700,000 from roughly $400,000 a year ago. Citigroup sees AI replacing 54% of all jobs in banking.