You are not inconsistent. You are two businesses.

Because you are doing two jobs that compete for the same hours. Selling and delivering cannot both happen in the hour you have. A ten-person firm has people who do only one of the two; you do both, alternately, and the alternation is what produces the oscillation. Swiss data puts the cost of that structure at 18.2 percentage points of five-year survival.

Which is to say the problem has a size, and the size is not a character trait.

The mechanism, stated plainly

An hour has one use. Spend it writing a proposal and it is not spent delivering. Spend it delivering and it is not spent writing a proposal. That is not a productivity observation; it is a constraint on a scarce input with no substitute.

Now add the delay, because the delay is what makes the cycle a cycle. Selling done today produces revenue in six weeks, three months, six months — whatever your sales cycle happens to be. Delivering done today produces revenue on its own schedule; in legal practice, the average total lockup between doing the work and holding the money is 93 days.

So the two activities pay out on different clocks, and the one you stop doing is always the one whose absence you will not feel for a quarter.

The rest follows mechanically. Busy period: delivery crowds out selling, because delivery has a deadline attached and selling does not. One sales cycle later: the pipeline is empty. Empty period: selling crowds out nothing, because there is nothing to deliver, so the pipeline refills. One sales cycle later: busy. The oscillation has a period, and the period is your sales cycle. It repeats because the constraint repeats.

You are not inconsistent. You are one person being asked to run two businesses with one person's hours.

What founding alone costs

The Bundesamt für Statistik tracks survival rates of newly founded Swiss companies. The figures below come from the 2013 founding cohort, on a page last updated on 7 December 2020 — and I want to date that explicitly rather than imply it is current. There is no newer BFS publication of five-year survival by size class.

Five-year survival of Swiss companies by size at founding, 2013 cohort. Bundesamt für Statistik, page last updated 7 December 2020.

Size at founding	Still trading after five years
1 person	47.3%
2–4 employees	57.7%
5–9 employees	60.4%
10 or more employees	65.5%

47.3% of Swiss companies founded by one person were still trading five years later, against 65.5% of those founded with ten or more employees — a gap of 18.2 percentage points.

Overall, across all sizes: 83.9% survive one year, 70.9% two years, 61.7% three years and 49.2% five years.

Now the caveat, because the number is not as clean as it looks. This is a difference between groups, not a controlled experiment. Companies founded with ten employees had capital, backers, and often a customer before they opened the door. Size at founding correlates with a dozen other things.

But notice what every candidate explanation has in common. Capital. Sector. Backing. Division of labour. Every one of them is structural. Not one of them is discipline. Whatever mix of causes produces those 18.2 points, personal consistency is not on the list of plausible contributors — and it is the only explanation most solo owners have ever considered.

Sector matters too, and more than most people assume.

Five-year survival by sector, Swiss companies founded 2013. Bundesamt für Statistik, page last updated 7 December 2020.

Sector	Five-year survival
Health and social	63.5%
Education	53.0%
Transport	51.9%
Professional, scientific and technical services	49.7%
Information and communication	46.7%
Arts and entertainment	45.7%
Finance and insurance	45.0%
Retail	42.2%
Hospitality	38.1%

Professional, scientific and technical services — the consultants, engineers, architects and analysts — sit at 49.7%, almost exactly the all-sector average. Health and social at 63.5% is the strongest line in the table, which is worth knowing if you run a therapy or clinical practice: the sector is not what is working against you.

The same gap turns up in the hours

If the survival figures were the only evidence, you could dismiss them as an artefact of capital. They are not the only evidence, and the second and third measurements come from different countries and different methods.

SPI Research's 2025 Professional Services Maturity Benchmark, covering 2024 data from 403 firms, measures billable utilisation at 64.3% in firms under ten employees and 78.3% in firms over 700. Fourteen percentage points, in the same direction.

The RIBA Business Benchmarking Report 2023 measures UK architecture practices and finds the sharpest version. Partners, directors and sole principals bill 46% of their time. Salaried architects, technologists and assistants bill over 70%, rising above 80% in practices of 100 staff or more. The same report puts overall practice profit margin at 2%.

Three measurements: survival, utilisation, billable share. Three countries, three methods, three separate research organisations with nothing to do with each other. All showing the same gap between people who do one job and people who do two.

Which is why the pipeline is empty exactly when you need it

The Freelancer-Kompass 2026 surveyed 5,412 freelancers across the DACH region. 43% have no secured project utilisation for the coming months at all. Among those who do: 12% are secured for up to one month, 20% for two to three months, 13% for four to six.

And 24% had fewer than 50 project days in the past year. Fifty days is ten working weeks. Roughly a quarter of an experienced professional population — the same survey reports an average of 18 years' experience — billed fewer than ten weeks of project work across twelve months.

That is not a distribution of skill. Skill does not oscillate on a quarterly cycle.

Why "be more consistent" is advice that cannot work

Consistency advice assumes the limiting factor is intention. Intention is not scarce here. Hours are.

You cannot be consistent at two activities that consume the same input and pay out on different schedules, in the same way you cannot spend the same franc twice by being more organised about it. The instruction does not fail because people ignore it. It fails because it is addressed to the wrong variable.

I have never met a solo expert who did not already know they ought to be selling while they are busy. Knowing it has never been the constraint. The hour is spent, and it was spent on the thing with a deadline and a client's name on it, which is the correct decision every single time it is taken and disastrous in aggregate.

If you have privately concluded that the problem is you, the evidence does not support it. Firms that began with employees survived at 65.5% against 47.3%. Firms under ten people bill 64.3% of their time against 78.3%. Sole principals bill 46% against over 80%. That is what a structural constraint looks like when three different people measure it without coordinating.

What the survival table does not say

It measures size at founding. It does not follow that a solo practice hiring in year four moves to 65.5%. Those are two different claims and the second does not follow from the first — the table has nothing whatever to say about what happens when a company changes size. I would rather be exact about that than sell you a headcount.

What the table does support is narrower and more useful: firms in which the two jobs were separated from the beginning survived more often, billed more of their time, and showed up better on every measurement anyone has taken. Separation is the variable that all the evidence touches. Hiring is one way to achieve it. It is not the only one, and it is the most expensive.

The two honest options are separating the jobs in time — a selling commitment with a structure around it that survives a busy month — or separating them in person, which can mean an employee, but can equally mean a delivery subcontractor, an associate who takes overflow, or a partner who does only one half.

Which of those fits depends on facts about your revenue: how long the cycle actually runs between first contact and first invoice, how lumpy the projects are, what share of delivery genuinely cannot be done by anybody else, and how much of the year you can afford to spend at 46% utilisation. Those are not knowable from the outside, and anybody who tells you which one you need without having seen them is guessing.

One warning about separating in time, since it is the option most people reach for first. A fixed selling commitment in a solo practice is enforced by the same person whose delivery deadline is competing with it. In a firm, a different person holds it, and that is the entire difference. "I will be disciplined about it" is the one enforcement mechanism that has already been shown not to work — not because of any deficiency in the person, but because the craft always wins the hour.

Solo is not worse at everything

Eurostat's analysis of the 2017 EU Labour Force Survey ad-hoc module, covering the EU-28, found 33 million self-employed people, 71.8% of them with no employees at all. This is the normal condition, not a marginal one.

18.2% had one dominant client — an unrelated 18.2, and worth keeping separate from the survival gap above. That figure is a concentration risk, and it compounds the oscillation: one client ending is a pipeline emergency and a delivery vacuum in the same week, which is also why the arithmetic of a price rise looks different for a concentrated book.

And on satisfaction: 46.1% of solo self-employed reported high job satisfaction, against 53.3% of self-employed with employees. A gap of 7.2 percentage points. That is a difference, not a catastrophe — nearly half of solo self-employed people report high satisfaction, which is not the profile of a population in distress. The data is from 2017 and covers the EU-28, so it predates a great deal, and satisfaction is self-reported. What it adds is one more measurement pointing at structure rather than at temperament.

What this means in practice

The oscillation has a period, and the period is measurable — which is the most useful thing in this article. The measurement is a plot of revenue by month against the date of first contact for each piece of work; the lag between the two peaks is your sales cycle. It is usually longer than people expect, and it is the single number that determines how far ahead selling has to run to keep delivery full.

Once that number exists, the question changes shape. It stops being "why can't I be consistent" — which has no answer — and becomes "how many weeks ahead does selling have to run, and what does it cost to protect that many weeks". The second question has an answer, and the answer is usually a smaller number than hiring.

It also reframes what a busy period is. A quarter at full delivery with nothing sold is not a good quarter. It is a good quarter followed by a bad one, and the bad one has already been decided by the time you notice it. The utilisation arithmetic and the pipeline arithmetic are the same problem seen from opposite ends of the table.

What this cannot tell you from the outside

The Swiss survival data is a 2013 founding cohort last refreshed in December 2020, and it measures size at founding rather than current size. SPI measures firms large enough to run time-recording systems, which most solo practices are not. RIBA covers UK architecture. freelancermap covers the DACH region, weighted towards IT and consulting. Eurostat's satisfaction figures are from 2017 and cover the EU-28. Not one of these is a benchmark for your practice.

What none of them can tell you is the shape of your own revenue — the real length of your cycle, what share of delivery only you can do, what a lost quarter costs you, and whether your book is concentrated enough that one departure is an emergency. All four are recoverable from documents you already have: invoices, the calendar, and the dates on your proposals. It is about a fortnight of unglamorous reconstruction.

It is also nobody's job. Your fiduciary reports the year that closed, as a single figure with the seasonality averaged out of it. Your bank sees the balance. Nobody you pay plots your revenue against the date of first contact — and that one chart is where the period of the cycle becomes visible, which is to say where the thing you have been calling a personal failing turns out to have a wavelength.