Notice which budget line is open before you design a solution, because the open line will decide it for you.
Why do companies freeze headcount but approve consultants?
Because the two costs are counted differently and one of them is scrutinized.
Headcount is permanent, visible, and carries obligations. Approving a position means committing to salary, benefits, and an ongoing line somebody will ask about every year. It is reviewed at the top of the company.
Services spend is a project cost. It has an end date, it does not appear in the headcount number the board looks at, and it is approved at a lower level.
The two are frequently comparable in money and treated as entirely different things. Nobody planned that. It emerged from how organizations count.
What happens when a company grows and the IT team does not?
I had eight people, and that was all I had.
The company grew. Systems were added, locations were added, volume increased. My staff did not grow with any of it, apart from very rare exceptions.
We were allowed to use consultants. So I used consultants, and that is why I hired an outside firm as a backstop for my database administrator, and why the same firm ended up running the private cloud, the computer room, and the overnight desk.
None of those were decisions about the best way to organize technology work. Each was a decision about which budget line was open.
How do you run infrastructure with a small permanent team?
Everyone owns their systems completely.
Each person was responsible for their own systems, and for the disaster recovery counterparts that went with them. Also the security of those systems. Also the performance. Also everything else in their area.
The person responsible for the middleware owned the application, the machine, the security, the performance, and the database underneath it. All of it, one person.
People learned a lot about a little and a little about a lot. That is what eight people covering a large estate produces, and it was not a philosophy. It was arithmetic.
Does full-stack ownership work in practice?
Full-stack ownership worked for us, and it was forced on us.
One person owning the application, the machine, the security, the performance, and the database is what the industry spent the following two decades naming and selling as a philosophy.
We arrived at it because we had eight people and no prospect of a ninth. Nobody read a book about it.
The benefit is real and it is not what the consultants sell. A person who owns everything about a system understands it in a way that four specialists sharing it never will. There are no handoffs, no seams, and no arguments about whose problem a symptom is.
The cost is that they cannot take a vacation, which is a separate problem and the reason the rotation and the cross training had to exist.
What does the consultant loophole cost long term?
Capability migrating out of the company, one reasonable decision at a time.
Each individual choice was correct. The work had to happen, the hours did not exist, and services spend was available while headcount was not. I would make the same calls again.
The aggregate is something nobody decided. Over several years the company came to depend on outside firms for capabilities it could not perform itself, and there was never a meeting where that was chosen.
That pattern is not unique to my situation. It is what a headcount freeze combined with an open services budget produces in almost every company, and the mechanism is invisible because each step is defensible on its own.
Should a small company hire or contract?
Ask which one you would choose if both budgets were the same.
That question separates the real decision from the accounting one. If contracting is the better answer for the work, contract. If you would hire but cannot, you have discovered that your organizational shape is being set by a budget rule rather than by the work.
Knowing that does not always change the outcome, because the constraint is real and arguing with it usually fails. What it changes is whether you see the drift coming.
A leader who knows the pattern can decide which capabilities to defend. Some things can go outside permanently and it costs nothing. Others are close to how the business works, and once they leave, the vendor understands your operation better than you do.
How do you argue for headcount when the answer is always no?
Show what the services spend already costs, side by side.
Most companies with a headcount freeze have no idea what they are spending on outside help across a year, because it arrives in separate approvals under separate projects. Adding it up is often the entire argument.
Then name what the money is not buying, which is permanent knowledge inside the building. That framing works better than workload, because workload arguments sound like complaints and capability arguments sound like planning.
It will still frequently fail, and the freeze will hold. Making the total visible at least means the tradeoff is being chosen rather than drifted into.
