Klarna – A B2C Case Study in Caution

Klarna’s IPO was supposed to be one of the interesting ones we’ve seen in quite a while.
This is why I was very enthusiastic when I started reading its prospectus. Alas, a few pages later, my enthusiasm turned into skepticism.

Why B2C Numbers Lie

B2C numbers are very tricky to interpret—not just as an investor. Even for internal data teams, interpreting them requires a lot of business understanding, extremely careful handling, and almost always: never relying on a single number.

The reason is simple: while in the B2B world, most customers will be of the same order of magnitude in lifetime value (or any other $$ metric), in B2C, the variation is huge. You’ll often find companies where 90% of revenue relies on just 0.5% of the clientele. Pareto—taken to the extreme.

This is why, when you read in a B2C filing that “no single customer contributes more than 10% of the revenue,” it means nothing. That’s a useful metric in B2B where there might be one or two very large customers. In B2C, it’s the Pareto ratio that reflects revenue-concentration health. Is it a 0.5–90? A 10–90?

But they usually don’t disclose that.

This is also why simple averages often conceal more than they reveal. Imagine a company with a million clients: 995,000 of them contribute just $1 each, and the remaining 5,000 clients generate $10M in total. The average revenue per customer is $11—so it might seem reasonable to spend $3–4 acquiring a new customer via marketing.

Or does it? If most customers generate a negative net value?

This is just a simplified example to illustrate why decision-making in B2C is super complicated, even when you have all the data in front of you.

What’s Up With Klarna

Klarna’s numbers are hard to interpret as is. But this is exactly why I suspect they’re either misinterpreting their own data (which is very easy to do in B2C), or presenting it selectively.

Before listing the issues, I’ll emphasize: each issue on its own is not a big deal. The problem is the pattern. Too many numbers are presented in ways that raise questions.

Let’s start:

• Active Users – 93 million.
  PayPal reports 434 million. So, is Klarna just 4x smaller than PayPal?
  Not quite. They use different definitions. PayPal only counts users who transacted in the period. Klarna counts anyone who logged in, even without transacting. Quite a relaxed definition of “active.”

• Underwriting Capabilities.
  Klarna highlights its proprietary ML models for underwriting loans, claiming this enables growth while keeping losses low. But when comparing its models to others, they show only a single best-case example—best model, best country. No averages, no cross-country breakdowns. Feels like a cherry-picked case.

• Small Net Profit.
  In Klarna’s initial prospectus, they showed a small 2024 profit of $21M, less than 1% of revenue or assets. Numbers that small deserve a second look, as even minor accounting changes can flip them. One such tweak: Klarna lowered its discount rate used for goodwill impairment testing—from 15.7% to 12.5%. If the company’s risk profile hasn’t materially improved, this change is questionable. And it directly inflates the recoverable value of assets, thus helping the bottom line.
  When profits are razor-thin, every adjustment matters.

• Partial NPS Data.
  Net Promoter Score (NPS) is the most common metric used by B2C companies to gauge customer satisfaction. It’s based on a single (and debatable) question: “How likely are you to recommend X to a friend?”
  It’s volatile, heavily influenced by the sample population, and generally only useful as a self-comparison over time, not across companies.

  Yet, despite presenting all KPIs in yearly terms, Klarna chose to show off a single September NPS score, claiming it’s higher than competitors’. That’s cherry-picking at best.

There were other eyebrow-raising items too—including some potential number misinterpretations (always in Klarna’s favor). But this was already enough for me to decide: Klarna wasn’t the exciting IPO I had hoped for.