Follow me

The Brand Strategy Gap Behind Klarna’s AI Mistake

The Brand Strategy Gap Behind Klarna’s AI Mistake

The European buy-now-pay-later market is built on a single fragile asset: trust. Not technology, not credit lines, not merchant partnerships. “Trust”… because the entire proposition asks a financially stretched consumer to believe that this company is on their side when they defer a payment. When you build a brand strategy for a business on people managing money anxiety, the customer relationship is the product. Everything else is infrastructure. That is the lens through which the Klarna story should be read and it is a lens most of the coverage missed entirely.

For anyone evaluating fractional CMO leadership or trying to understand where senior brand strategy actually earns its value, Klarna is one of the most instructive corporate case studies of the last two years. Not because it failed – it didn’t but because of what its correction reveals about a gap sitting inside almost every modern organisation.

What Klarna Actually Did And What It Walked Back

In February 2024, Klarna announced that its OpenAI-powered assistant was handling the work of 700 customer service agents. CEO Sebastian Siemiatkowski went further than the technology itself, telling Bloomberg he believed AI could already do all the jobs humans do. The company implemented a hiring freeze that lasted over twelve months and framed its shrinking headcount as a model for every services business to follow. The tech press treated it as proof the AI productivity dividend had finally arrived.

Fifteen months later, the message changed. In May 2025, Siemiatkowski returned to Bloomberg with a different tone: “We went too far. We focused too much on cost. The result was lower quality.” Klarna began reintroducing human agents. The headline read as a reversal: AI experiment fails, humans return.

But the precise story is more useful than the headline. Klarna did not shut down its AI. By Q3 2025, the assistant was doing the equivalent work of 853 agents, which is more than at launch. With roughly $60 million in annual savings and response times 82% faster than the pre-AI baseline. What Klarna corrected was not the technology. It was the scope. The company had allowed AI to absorb customer moments it was never suited to handle: disputes, fraud claims, account escalations, the emotionally charged conversations where a financially anxious customer most needs a human.

Klarna optimised for the average ticket, not the dangerous one. That is not a technology failure. It is a brand architecture failure.

The Numbers Behind the Correction

The data tells a more nuanced story than either the initial hype or the reversal narrative suggested. AI volume actually grew through the correction period, what changed was where the humans sat within the model:

Metric At Launch (Feb 2024) After Correction (2025–26)
Work handled by AI assistant 700 agents’ equivalent 853 agents’ equivalent (Q3 2025)
Total headcount ~5,000 (end 2022) ~3,422 (end 2024)
Annual cost saving claimed ~$40M projected ~$60M reported (2025)
Response time vs pre-AI Baseline 82% faster
Human support strategy Freeze – AI to replace Hybrid – humans for complex & premium
CEO public position “AI can do all our jobs” “We went too far… lower quality”

 

The Reframe: This Was Never an AI Problem

The easy read is to call this a failed AI experiment. That read is wrong, and it misses the lesson every marketing leader should actually take from it. Klarna didn’t have an AI problem. It had a brand architecture problem.

Nobody in the organisation was holding the line on which moments were brand-critical and which were genuinely automatable. The decision to automate customer service travelled from a cost mandate through operations and technology, the people who could implement AI. It never passed through anyone whose job was to define what the Klarna experience must feel like regardless of how it is delivered. So AI was allowed to optimise everything, including the one touchpoint where the brand promise lived.

For a BNPL company, this is not a minor error. The sector is under active regulatory scrutiny across the EU and UK precisely because it operates in the space of consumer financial vulnerability. Trust is the entire licence to operate. Automating the moment of money stress does not just hurt service metrics; it contradicts the brand promise at the exact point the customer is testing whether that promise is real.

The Questions Klarna Skipped

The questions that should have been asked before the automation decision are the ones I find most instructive. They are not technology questions. They are brand strategy questions, and they belong to whoever owns brand strategy at the most senior level:

  • Which touchpoints carry the brand promise and must never be fully automated?
  • Where does AI amplify what we stand for, and where does it quietly erode it?
  • Who in this organisation has the authority to say no to an efficiency decision on brand grounds?

That last question is the real gap and not just at Klarna. Across most of the market, AI adoption decisions are being made by people optimising for cost and capability, with no one senior enough in the room whose mandate is to protect the brand-critical human moments. The person who should define where automation ends is typically absent from the conversation in which automation is decided.

The person who should define where automation ends is typically absent from the conversation in which automation is decided.

Why This Is the Strongest Case for a Fractional CMO

This is where fractional CMO leadership earns its value, not as a cheaper full-time hire, but as a structurally different kind of authority. A fractional CMO sits outside the cost-cutting pressure. They have no internal promotion incentive to champion the efficiency narrative that impresses the board. They have no political reason to avoid telling operations that a particular automation decision will damage the brand. Their entire value lies in their ability to hold the line on brand-critical human moments precisely because they are not part of the machinery generating the pressure to automate them.

The fractional model works here for the same reason it works in strategic continuity generally: the person is senior enough to have the conversation at board level, external enough to be honest about it, and focused enough to define the human-AI boundary before the efficiency decision is made rather than after it backfires. That timing is the whole game. A correction is not a strategy. Reacting to a mistake is not the same as having a framework that prevents the next one.

I want to be precise about what I am and am not claiming. I do not have Klarna’s answer, and nobody credible does from the outside. The specific boundary between what Klarna should automate and what it should protect depends entirely on being inside that business, understanding its customers, and working through the trade-offs with the people who own them. What I can say with confidence is what question they skipped and that it needed to be asked before the automation, not after the reversal.

The Takeaway for Every Brand Adopting AI

Klarna is not a cautionary tale about AI. It is a cautionary tale about who is in the room when AI decisions get made. Every brand currently under board pressure to demonstrate AI adoption is at risk of the same gap, not because the technology is dangerous, but because the strategic authority that should define its limits is missing from the conversation.

If your organisation is making AI-integration decisions right now, the most valuable question you can ask is not “what can we automate?” It is “who here is responsible for deciding what we must never automate and are they actually in this meeting?” If the answer is no, that is the gap. And it is a gap worth closing before the correction, not after it.