The design problem nobody is talking about

When we write about enterprise technology at Sigma, the focus is usually on the systems that face end users: the checkout flows, the government portals, the workplace applications that waste 112 hours per worker per year. But there is a category of enterprise UX failure that operates almost entirely out of view — and it is arguably the one that destroys the most value in early-stage companies.

It is the design of the financial back office.

For a founder building a SaaS company, an e-commerce brand, a fintech platform, or any kind of modern technology business, the infrastructure that processes invoices, reconciles bank feeds, handles payroll, and files tax returns is not glamorous. It is not the product. It does not appear in investor decks. But research consistently shows that poor financial system design is one of the leading causes of stress, cash flow surprises, compliance failures, and — in the worst cases — company death.

According to the Bank of England’s 2024 report on AI in financial services, around 75% of UK financial services firms are already using AI to improve back-office workflows. Mordor Intelligence projects that the global AI accounting market will reach $10.87 billion in 2026, with small and medium-sized enterprises driving adoption at a 44.6% compound annual growth rate. The shift is dramatic. But it is also unevenly distributed. The firms that have gotten ahead — the ones who have restructured their back office around modern, AI-native tools and expert human oversight — are pulling away from the rest.

To understand what separates the companies that treat their finance function as a design problem from those that treat it as a cost center, we spoke with Daniel Martinsson, Partner at Sveago, a Swedish accounting and legal firm that specialises in working with technology founders. Martinsson has spent the last decade watching the quiet collapse of countless startups whose product was strong but whose financial architecture was fragile. His perspective on the state of founder finance in 2026 is, by turns, sobering and pragmatic.

Part 1: The interview

Sigma: Let’s start with the obvious question. Why do smart, technically capable founders get their back office so wrong?

Daniel Martinsson: It’s almost always the same story, and it’s almost never about intelligence. It’s about attention. When you’re building a product, your attention is finite. You spend it on customers, on code, on hiring. Everything else gets deferred. And the back office is the easiest thing to defer because, for the first 18 months, nothing goes visibly wrong. The invoices go out. The money comes in. You file taxes in April. Life is fine.

Then you hit around 20 employees, or you raise a proper round, or you start selling into three countries, and suddenly everything you deferred arrives at once. You have a board that wants monthly reporting. You have a tax authority that wants transfer pricing documentation. You have an investor who wants to understand your cash runway by cohort. And you’re trying to build all of that on top of a spreadsheet and a bookkeeper who comes in on Thursdays.

That’s the moment most founders realise that the back office was never a cost center. It was a design decision they made by not deciding.

Sigma: You work specifically with technology founders. What are the patterns you see in how they set up their financial systems?

Martinsson: There are three archetypes, and they all fail differently.

The first is what I call the DIY founder. They’ve read about Fortnox, they’ve watched a YouTube tutorial, they think accounting is just data entry. They do their own books for 12 months, it works fine because the company is simple, and then they try to close their first proper bokslut and discover that three quarters of their transactions are miscategorised. Fixing it costs more than hiring a professional from day one would have.

The second is the delegated founder. They hire a cheap bookkeeper, usually a family friend or someone recommended in a Slack channel. The bookkeeper is competent at data entry but has no perspective on what the founder is actually building. When the company needs advice on 3:12 optimisation, on how to structure founder compensation, on when to move from K2 to K3 — the bookkeeper shrugs. The founder gets technically accurate books that are strategically useless.

The third is the over-engineered founder. They’ve hired a Big Four firm because they saw another founder mention it on LinkedIn. They’re paying 15,000 SEK a month for services they don’t need, because the firm was built to serve 500-person companies, not 8-person startups. The relationship feels prestigious but the actual service is distant and generic.

What all three have in common is a failure of design. They haven’t asked the right question, which is: what does a finance function actually need to do for a 2026 technology company, and what’s the minimum sustainable structure that can do it well?

Sigma: Let’s dig into that. What does a modern finance function need to do for a growing tech company?

Martinsson: At the most basic level, six things. Bookkeeping, payroll, tax, reporting, compliance, and advisory. But the weighting of those six has shifted enormously in the last five years.

Five years ago, maybe 70% of a founder’s accounting spend went to bookkeeping and tax filing — the purely mechanical stuff. Now, if you’re doing it right, that number should be closer to 30%. The rest should go to advisory, reporting, and compliance design. Why? Because AI has collapsed the cost of the mechanical work. What used to take a bookkeeper four hours now takes software 40 seconds. The question is no longer “who can do the data entry” but “who can interpret the data once it’s entered.”

Part 2: The numbers behind the shift

Martinsson’s observations align with a growing body of research. The consistent finding is that AI has not replaced accountants — it has redefined what accountants do. Meanwhile, the cost of getting the back office wrong has risen, not fallen, because the complexity of the regulatory environment continues to increase.

Consider the following data points, drawn from recent industry research:

• 99% accuracy for AI-powered invoice processing, versus 85–90% for manual data entry. Each manual error costs an average of $53 to correct, before compliance penalties.
• 60–80% reduction in payroll processing errors when AI validation is layered onto traditional payroll workflows.
• Monthly close cycles have shrunk from an average of 11 business days in 2020 to 4 business days in 2026 for firms using modern tooling.
• 30–45% improvement in forecasting accuracy for companies that have replaced static annual budgets with AI-driven rolling forecasts.
• 65% reduction in tax preparation time for firms using AI-assisted tax workflows, with over 80% of individual tax return preparation now automatable.

These numbers paint a picture of an industry in the middle of a quiet productivity revolution. But the benefits accrue only to the firms — and the founders — who have actually invested in the redesign. The rest are being left behind, paying 2021 prices for 2021 service levels while their competitors pull ahead.

Sigma: This sounds almost too optimistic. What are the downsides?

Martinsson: There are real ones. The first is data quality. AI tools are only as good as the data you feed them. If your bank feeds are broken, if your invoicing system is disconnected from your GL, if your employees are still submitting expense receipts on paper — all the AI in the world won’t help you. The upstream data architecture matters more than the downstream automation.

The second is judgment. AI is excellent at processing and pattern recognition. It is bad at understanding context that isn’t in the data. If you’re making a call about whether to capitalise a cost or expense it, about how aggressively to interpret a tax rule, about whether a transaction is truly at arm’s length — these are judgment calls that require a human who understands your business and your risk tolerance. Founders who think AI will handle this are setting themselves up for very expensive surprises.

The third, and this is the one I worry about most, is what I’d call illusory transparency. Modern dashboards look beautiful. Real-time cash flow, colourful charts, everything at your fingertips. But dashboards create the feeling of control without the substance of it. If nobody on your team actually understands what the numbers mean, the dashboard is just wallpaper. You need the human layer — someone who can say “this number looks fine but here’s why I’m worried about it” — or the technology is just decoration.

Part 3: The Swedish tech scene and its peculiar geography

Sigma: Sweden punches well above its weight in tech. Spotify, Klarna, King, Mojang, iZettle, Truecaller, Northvolt — the list is extraordinary for a country of 10 million people. Is the finance function contributing to that or holding it back?

Martinsson: Both. Sweden has unusually founder-friendly tax and company law for early-stage companies. The 3:12 rules, for all their complexity, are actually quite generous to founder-shareholders who structure things correctly. The fåmansföretagsregler create real opportunities for tax-efficient compensation. ROT and RUT deductions benefit certain business models. A founder who understands the Swedish system can extract more value than they could in many other European jurisdictions.

The flip side is that if you don’t understand the system, it punishes you brutally. I’ve seen founders pay millions in unnecessary tax because they structured their cap table wrong in year one and couldn’t undo it in year five. I’ve seen founders miss the qualifying holdings rules and lose their preferential dividend treatment. I’ve seen founders ignore the K10 form for three years and then try to catch up in the middle of a due diligence process for a sale. These mistakes are incredibly expensive and almost always preventable.

Sigma: Is there a geographic dimension to this? Where are the Swedish tech founders actually clustering?

Martinsson: Stockholm still dominates, obviously. But within Stockholm the picture has shifted in the last five years. The traditional tech cluster was around Kungsholmen and the central districts, close to the old media and finance firms. Now you see real density in the northern suburbs — Kista of course for the larger tech companies, but also Solna and Sundbyberg for fintech and B2B SaaS, and increasingly Sollentuna for a kind of quiet, under-the-radar founder community that prefers lower costs and better family infrastructure.

Sollentuna is interesting specifically because it’s become a base for founders who don’t want the noise of central Stockholm but still want to be 15 minutes from Arlanda and 20 minutes from Stureplan. There’s a growing cluster of smaller SaaS companies, consultancies, and tech services businesses there, and the local support infrastructure has started to catch up. If you want an accountant who actually understands what it means to run a fast-growing SaaS business and can meet you in person without requiring a two-hour commitment, a local redovisningsbyrå i Sollentuna is now a viable alternative to the Stockholm city firms, and often a better one for the 5 to 25 employee segment where most growth happens.

But the geography matters less than it used to. The real question is whether your accountant understands software businesses. That skill is distributed unevenly across the country.

Part 4: Practical guidance for founders

Sigma: Let’s get tactical. If you were sitting down with a founder of a 12-person SaaS company in Stockholm tomorrow, what would you tell them to do about their finance function in the next 30 days?

Martinsson: Five things.

One: Audit what you have. Not the books — I mean the system. Who does what, how long does it take, where are the handoffs, where are the bottlenecks. Most founders have never actually mapped their finance workflow. Once you map it, the problems become obvious.

Two: Calculate your real cost. Not what you pay the bookkeeper — what you and your team spend on finance-related admin every week. The hidden labor cost is usually 3–5x the explicit cost. Founders are often shocked by this.

Three: Identify the three things that cause you the most anxiety. Is it cash flow visibility? Is it the month-end scramble? Is it not knowing what your tax bill will be? Whatever those three are, they’re what your new system needs to solve. Everything else is secondary.

Four: Don’t start by changing software. Start by changing structure. The most common mistake I see is founders who think the problem is their tool, when the problem is their process. Tools are easy to change. Processes take longer. Fix the process first.

Five: Get a second opinion. If you’re with a firm that isn’t meeting your needs, don’t just complain about it — talk to someone else. You’ll learn more in a 30-minute conversation with a different accountant than you will in six months of frustration with the one you have.

Sigma: What about the founders who say “I’m too small for all of this, I just need cheap books”?

Martinsson: I hear this a lot and I think it’s a mistake, but I also have sympathy for it. When you’re four people and bootstrapping, every krona matters. You don’t want to pay 8,000 SEK a month for accounting services when you’re not sure if you’ll still be in business in six months.

My honest answer: at that stage, yes, use the cheapest competent solution you can find. Fortnox plus a freelance bokförare is fine. But three things will change the math, and you need to watch for them.

First, if you take outside money. The second a VC writes you a check, your reporting requirements go up by an order of magnitude and a freelance bokförare is no longer sufficient.

Second, if you hire beyond six people. Payroll complexity scales non-linearly. At six people it’s manageable. At ten it’s a real job. At fifteen you need proper systems.

Third, if you start selling internationally. VAT rules across EU borders are an actual nightmare, and doing them wrong creates retroactive liabilities that can ruin a small company.

If any of those three things happens, upgrade your finance function within 60 days. Don’t wait.

Part 5: Looking ahead

Sigma: Where do you think the industry is going in the next three years?

Martinsson: Three predictions.

First, the middle of the market will hollow out. The Big Four firms will continue to serve large enterprises, and a new generation of AI-native boutique firms will serve the SME market. The mid-sized traditional accounting firms — the ones who charge premium prices for commodity work — will struggle. Their business model was built for a world where data entry was expensive. In 2026, it isn’t.

Second, the advisory function will separate from the compliance function. Today, most firms bundle everything together. In the future, I think you’ll see founders using one provider for mechanical compliance work (often fully automated, very cheap) and a separate advisor for strategic decisions (human, more expensive, much higher value). The bundling will break.

Third, the geographic center of gravity will shift. Right now most premium accounting services are concentrated in city centers. As remote work normalises and digital delivery improves, you’ll see more serious firms operate from suburban locations, where costs are lower and clients are closer. This is already happening in Stockholm. It will happen everywhere.

Sigma: Any final advice for founders reading this?

Martinsson: Stop thinking about accounting as a cost center and start thinking about it as an operating system. Your finance function is the substrate on which every other decision runs. If the substrate is broken, everything on top of it is unstable, no matter how well-designed. If the substrate works, you can move fast and take risks with confidence, because you know the numbers are telling you the truth.

The founders who get this right are not the ones who spend the most. They’re the ones who think clearly about what they need, who choose partners based on fit rather than prestige, and who treat their back office with the same design discipline they apply to their product. That’s not a huge ask. But it’s a lot more than most founders currently do.

Editorial note: the design lesson

At Sigma, we spend a lot of time writing about the design of systems that users see. The checkout flows, the government forms, the workplace applications. Martinsson’s perspective is a useful reminder that some of the most consequential design decisions in any company happen in systems users never see at all.

The finance function of a growing company is, in the end, a user experience problem. The user is the founder. The job is to give them accurate information, on time, without friction, so they can make good decisions. The failure modes are familiar: too much friction, too little information, confusing presentation, hidden errors, dashboards that look good but don’t reflect reality. These are the same UX failures we document in consumer applications every week. They simply happen in a domain that rarely gets the design attention it deserves.

For founders reading this: the next time you think about your back office, ask the same question you would ask about your product. Is this designed for the person who has to use it? If the answer is no — and it usually is — that’s where to focus.

The tools to fix it now exist. The expertise is available. The only question is whether founders are paying attention.