A project manager is starting with us on the 17th. She sent her contract review back this morning, and I’ve been sitting with it for about an hour. Most candidates sign and move on. She asked questions about three specific clauses. They were the three clauses I would have asked questions about. In twelve years of hiring, that has happened maybe four times.
The clauses she circled are the same three I now think are the company, the way the front of a building is not the building but is the part of the building people see.
The clause about working hours
The standard formulation in a Swedish salaried-employee contract goes something like: forty hours per week, plus what’s needed to fulfil your responsibilities. The first half is a number. The second half is a door left open. If you don’t qualify the second half, what you’re really signing is “unlimited hours when the company decides the responsibility requires it.” Most candidates don’t read that sentence carefully. The good ones do.
We added a sentence about symmetry. If a day runs eight and a half hours because the work that day required it, balance it with a seven-and-a-half-hour day later in the week. The symmetry is the point. The company is not enforcing the symmetry from the manager’s side; the employee is mostly the one keeping track. What matters isn’t the enforcement, it’s that the principle exists in writing, which means the company has said out loud that the balancing direction also counts.
There are agency contracts I’ve read where the balancing direction is absent. Forty hours per week, plus what’s needed, and that is the end of the sentence. Those contracts almost always come from agencies where the working culture has drifted in the direction the principle would have caught. Drift is easier than people think. The contract is the cheapest place to install a brake on it.
The clause about overtime
The standard formulation continues: overtime is compensated through your salary and other benefits. That sentence is technically fine. It’s also doing something the writer may not have noticed — it accepts overtime as a normal part of the job and then renders it invisible by saying it’s already paid for.
The version we use adds: requested overtime essentially does not happen here. Note the hedge. Not “never.” The agency that writes “never” into a contract has either never run a real client crisis or is being dishonest. “Essentially” is the load-bearing word. It says: there is a cultural floor; rare exceptions exist; we are not pretending the exceptions are zero.
What the hedge buys you is more interesting than the policy itself. A candidate reading “essentially never” knows two things at once. They know overtime is not the company’s lever. They also know that if a real exception happens, the company will treat it as the exception it is, rather than the unspoken expectation it was always going to be. The hedge is the honest part. Without it, the sentence is unenforceable, and a candidate who has read enough contracts can smell the unenforceability from a paragraph away.
In ten years of operating with that sentence in the contract, the count of explicitly requested overtime episodes across most multi-year tenures is somewhere between zero and a single digit. Not coincidence. The sentence shaped what we felt allowed to ask for.
The clause about when the salary lands
Most Swedish agency contracts pay salary the month after the month earned. So a person who starts on May 17th would get their first partial salary on June 25th. There’s a working-capital reason for the lag, and most accounting setups arrange for it by default. It’s also the place where the cost of changing the default is roughly nothing, while the signal of changing it is roughly large.
If you pay salary the month it’s earned — which means the May salary lands in late May — you are doing one of the cheapest things a company can do to communicate that it is optimizing for the employee’s cashflow as well as its own. It is, in absolute terms, a one-time working-capital shift of about a month, handled through the books. Accounting can do this. Our external bookkeeper raised an eyebrow the first time we asked and then arranged it without much further fuss. After that it was simply how we paid.
The signal does not depend on the size of the salary or the seniority of the role. A person in their first month of work, having just moved between agencies, often has the month’s gap as a real cashflow problem. Closing it costs the company a one-time line item and earns the kind of small, durable goodwill that does not show up in any quarterly review but does show up in retention. Compounding works in this direction too.
What we deliberately left out
You can write a hundred clauses into an employment contract and a candidate will read three of them. You can also leave clauses out, which is a separate kind of writing.
Non-compete clauses, for most salaried roles, we left out. Swedish law makes them broadly unenforceable for non-executive employees, which means writing one is a signal of distrust without the legal teeth to back it up. The signal is the only thing the clause does, and the signal is the wrong one.
Non-solicitation clauses on standard employees: also left out. Legally sounder, culturally colder. Either people leave on good terms and refer their networks back to you, or they leave on bad terms and the non-solicitation paperwork won’t stop them anyway. Skip it.
Bonus and variable compensation formulas embedded in individual contracts: left out for everyone below the partner tier. We had profit-sharing at the partner level and left it there. Cascading individual bonus formulas to project managers and specialists reliably teaches people to game the formula. The contract is the wrong place to install that incentive. If you want to reward a specific behaviour, fund a discretionary spot bonus and praise the behaviour directly. The contract should not be doing both salary and behavioural-economics duty.
Detailed project-allocation rules in the contract: also left out. Agencies that contractually pin a person to a vertical or an account end up either renegotiating contracts every six months or running an organization that can’t reallocate when the work demands it. Keep allocation operational, not contractual.
The pattern is the same in each case. A clause that’s mostly a signal of distrust is worse than its absence. A clause that locks in operating rigidity is worse than its absence. A clause that incentivizes gaming is worse than its absence. The contract should be doing the small set of things contracts are uniquely good at, and stopping there.
The job a contract actually does
A good employment contract is mostly empty. It needs to be legally complete, which Swedish standard templates handle. What it adds, beyond that, should be a small number of clauses that say something specific about how you intend to work together. Three or four. Not twenty.
The three above are the ones the new project manager asked about this morning. They are also the three I would have asked about if I were the one starting on the 17th. The reason that overlap happened is not, I think, that her preferences happened to match the contract by coincidence. It is that anyone reading a contract carefully reads the same three clauses carefully — the ones that touch how it actually feels to work somewhere. The rest is boilerplate. The candidate knows it; the company knows it; pretending otherwise is just noise.
What the contract does, at its best, is tell the candidate which battles you’ve already had with yourself and won. Working hours, with symmetry. Overtime, with an honest hedge. Salary timing, optimized in their direction. Those three say: we’ve thought about this, we’ve made it explicit, we are willing to be held to it. Everything else is paperwork.
The new project manager signed her version back this afternoon. I’m looking forward to the 17th. What I’m more interested in, frankly, is which three clauses in your agency’s standard contract are actually carrying the culture, and which three are the ones you wish you had drafted differently.
Hit me back.