Client numbers are falling for the first time. The growth mechanics are breaking.

There is one figure in the Norwegian accounting industry that has been almost boringly predictable until now: the number of clients has gone up. New companies are established, some drop off, and the net result is positive. That mechanic has made it possible to budget for growth without considering much beyond capacity.

That assumption no longer holds. According to figures and analysis from Regnskap Norge, as reported by Økonomi24, growth in assignment volume began to slow around 2023, turned negative in 2024, and the industry is now also losing clients, something the organisation describes as historically unprecedented (Økonomi24). That is two declines, not one. Both the volume per assignment and the number of assignments are pointing downwards at the same time. It is that combination that makes the figures interesting, and that makes "AI threat" too simple a headline.

The arithmetic behind an accounting firm's budget

Revenue in an accounting firm is, in practice, three numbers multiplied together: number of clients, volume per client, and price per unit. Historically, the industry has had ready access to all three. The first has grown on its own, the second has followed client activity, and the third has followed wage and price developments.

Now driver one is negative and driver two is negative. That leaves price as the only available lever. Here lies the uncomfortable detail in the figures: general price growth is itself cited as one of the reasons clients are dropping off, alongside the fact that small clients increasingly handle their own accounting through self-service solutions and AI models (Økonomi24). A price increase in the lower part of the portfolio therefore partly finances its own attrition effect. What remains as a genuine growth driver is revenue per existing client, not as a higher hourly rate for the same work, but as new content the client is actually willing to pay for.

Willingness to pay is a property of the client

This is where Capassa's reading diverges from the usual industry conversation. Advisory work is often discussed as something a firm decides to deliver. In practice, willingness to pay for advice is a property of the client, not of the firm. A limited company with flat revenue, one employee, no debt and no investment plans has limited need for decision support. A company that is growing, with loans, investments, falling gross margin, liquidity swings or an ownership transition ahead, has real questions with real financial consequences. Today, those two clients are often billed on the same principle: hours times rate. The difference between them is not visible in the timesheets. It is visible in their own financial statements. That is the shift the figures from Regnskap Norge really signal: when volume growth disappears as a budget driver, it is the clients' economics, not the firm's hours worked, that becomes the data foundation the budget has to be built on.

The average client does not exist

A budget that applies a uniform percentage growth in revenue per client across the entire portfolio assumes that the potential is evenly distributed. It almost never is. In a portfolio of a few hundred clients, a minority will have enough growth, complexity and capital requirements to carry a significantly larger fee. Another minority will be structurally on their way out, precisely the clients who, according to Regnskap Norge's review, are solving the tasks themselves (Økonomi24). Between them sits a large middle group where the question is standardisation and cost per assignment, not additional revenue. An average hides all three, and a budget built on the average allocates resources in the opposite direction from where the revenue can be found.

AI tools solve the cost side, not the revenue side

Svein Austheim, head of analysis at Regnskap Norge, argues that the industry should use AI to increase the value of its advisory work rather than fear automation, because the simplest tasks will increasingly be carried out by clients themselves (Økonomi24). Capassa's addition to that point is a clarification of what automation actually does to the arithmetic: it lowers cost per assignment. It does not create willingness to pay.

There is an asymmetry in the technology development worth noting. The tools in the client's self-service solution and the tools in the firm's production improve at the same time, but improvement on the client's side removes work from the firm, while improvement on the firm's side removes cost. Neither adds revenue by itself. A firm that becomes more efficient without changing its revenue model therefore ends up with a better margin per hour on a shrinking base of hours, an improvement that looks good in one year and problematic in three.

The sequence is the real question

Much of the industry discussion now concerns which tools to choose, but the figures suggest the sequence runs the other way. The segmentation of the portfolio determines which services have a market, which pricing model is possible, where capacity should sit and which competence needs to be recruited. If the tools are chosen first, without knowing which segments carry the revenue, what gets automated is the work already in hand, not the work that is supposed to pay for 2027. A budget built on revenue per client looks technically different from a volume budget: it does not consist of a single growth percentage, but of several groups with different assumptions, one group where the assumption is additional revenue, one where the assumption is lower cost per assignment, and one where the realistic assumption is attrition.

What the industry was actually told

2024 was the year a piece of arithmetic stopped working. That is a more demanding message than "technology is coming", because it cannot be answered with a purchase. The data needed to answer it already sits inside the firms, the clients' own financial figures. But it is usually organised for reporting per client and per period, not for analysis across the entire portfolio. Growth, margin, leverage, investment pace and liquidity are known client by client, and unknown as a distribution. That is the difference Capassa considers decisive once volume growth disappears, for the break in the growth mechanics is not primarily a technology problem, but an analysis problem, and it concerns the client portfolio a firm already has.