For the first stretch of a company’s life, buying computers is not a process. Someone joins, someone else orders a laptop on a card that evening, and it shows up at whatever address the new hire gave. It works because the whole thing fits in one person’s head.
Then a hiring plan lands, a handful of starts get scheduled in the same month, and the cracks arrive all at once. Nobody can say which machine belongs to whom. Two people have the wrong dock. Someone is waiting on a laptop that shipped to an old address. The founder who used to order them has better things to do and has quietly stopped replying.
You can usually date the transition to a single week, and it is almost always the week somebody tries to write the hardware situation down for the first time.
The moment buying stops being a personal errand
The signal is not headcount. It is the first time somebody asks a question that the card statement cannot answer. Which of these machines is out of warranty. Who has the one with the cracked bezel. How many spares do we have. What happens to the laptop belonging to the person who left in August.
Those questions mark the point where procurement becomes a small system rather than an errand, and the useful move is to build the system while the fleet is still small enough to fix cheaply. Twenty machines bought badly can be lived with. A hundred bought the same way becomes the thing your first IT hire spends their first two quarters undoing.
Specify by role, not by price tier
The common early mistake is picking one good machine and giving it to everyone, or picking three tiers by seniority. Neither maps to the work. What actually drives the spec is what the software does to the hardware all day.
| Role pattern | What actually drives the spec |
|---|---|
| Backend and platform engineering | Memory first, then sustained multi-core performance for builds and containers; internal storage headroom |
| Frontend and design | Colour accuracy and screen real estate; a second display more than a faster processor |
| Video, 3D and data science | Sustained thermal performance and discrete graphics or accelerator support; fast local storage |
| Sales, support and customer success | Camera, microphone, speakers and battery life; light chassis for travel |
| Finance and operations | Two displays, a comfortable keyboard, and total stability over peak speed |
| Shared and kiosk use | Serviceability, a wired connection, and a chassis that tolerates being moved |
Write those profiles down once and give them names. From that point on, hiring a backend engineer means ordering profile two rather than starting the same conversation from nothing, and the person doing the ordering does not need to understand the work to get it right.
The other advantage of profiles shows up at renewal. When every machine in the fleet belongs to a named profile, you can see at a glance which profiles are aging out and budget for them separately instead of treating the whole estate as one undifferentiated pile.
Standardise before you scale, because it gets expensive later
Standardising early is cheaper than it sounds, and the reason is supply rather than discipline. Buying one model in quantity, from a channel where that model still exists in quantity, is what makes a matched fleet affordable at an early stage, and it is a large part of why how refurbished equipment helps businesses scale tends to come up in conversations about hiring speed rather than conversations about cost. A machine that was sold to enterprises in enormous volume is available in matched batches for years afterwards. The newest release is available in ones and twos.
A fleet of eight identical machines behaves very differently from a fleet of eight different ones, and the difference has nothing to do with the machines. One model means one image, one set of drivers, one dock, one charger, one keyboard layout, one known set of quirks. When something breaks, you have seen it break before.
Mixed fleets generate a category of problem that looks like hardware failure and is not. The dock that works on four laptops and not the fifth. The display that will not wake on one generation of a chipset. The person whose charger is subtly different and therefore permanently missing.
Cash timing is the constraint, not the total
Early-stage buying decisions rarely turn on what something costs. They turn on when you have to pay for it, and on how wrong the hiring plan might be.
Three shapes come up repeatedly:
- Buy outright, in batches sized to the hiring plan you actually believe. Simplest to administer, hardest on cash in the month it happens, and it leaves you owning assets you will eventually have to dispose of properly.
- Lease or subscribe. Spreads the cost, keeps the refresh cycle on somebody else’s calendar, and commits you to a term whether or not the team grows into it. Read what happens at the end of the term before you sign the start of it.
- Buy a standing buffer, then replenish monthly. Keeps a small number of configured machines on a shelf so a start date never waits on a delivery. Ties up a little capital permanently and removes an entire class of onboarding failure.
The third shape is the one most fast-hiring teams end up at, usually after a new engineer spends their first week on a personal laptop they cannot install anything on.
Whichever shape you pick, the number that matters is not the unit cost. It is the cost of a person sitting idle, which is larger than any equipment decision and almost never appears in the comparison.
The onboarding kit is part of the purchase
A laptop on its own is not a working setup, and discovering that on somebody’s first morning is a bad way to discover it. The kit is the unit of purchase: machine, dock, cables, the right display or displays, keyboard and mouse if the person wants them, a headset that survives back-to-back calls, a lock if the office needs one, and an asset tag already applied.
Ordering those separately from different places, on different lead times, is how you end up with a complete laptop and no USB-C cable, which is a strange thing to explain to a senior hire on a Monday.
Where a supplier can do it, having machines arrive already enrolled in your device management, already imaged, already tagged and already recorded against a serial in your asset register is worth more than a discount. It converts onboarding from a half-day of setup into handing someone a box.
The day one of them breaks is a purchasing decision too
Nobody thinks about failure at the point of buying, which is why the first dead machine at a growing company is always handled badly. Someone loses most of a day, someone escalates it to a founder, and a replacement gets bought at full price under time pressure.
The fix is dull and it is decided at purchase. Keep a configured spare for every profile you run, or at least for the profiles where a day of downtime costs real money. Know before you buy who you contact when a unit fails, whether a replacement ships before the faulty one arrives back or after, and who pays the freight in each direction. Ask whether the supplier holds spare parts for the specific models they sold you, because a repair path that depends on sourcing a part from a marketplace is not a repair path.
Then agree what happens to the drive in a failed machine, because that question turns into a security question the moment a laptop with company data on it goes into a courier bag. Whether you keep the drive, destroy it yourself, or accept a certificate from the supplier is a policy call, and it is much easier to make calmly in advance.
None of that costs anything to arrange while you are still choosing. All of it is expensive to arrange in the middle of an incident.
Write down the boring things now
The asset register is the artefact that makes every later decision possible, and it takes an hour to start and years to reconstruct if you do not.
- Serial number, model, profile and purchase date
- Who currently holds it, and who held it before
- Warranty status and when the coverage ends
- Planned refresh date, so renewals stop being surprises
- Where the machine goes at end of life, and who wipes it
That last line is the one nobody writes down, and it is the one that gets asked about during a security review or an insurance renewal. A machine that leaves the company without a wipe record is an open question in your file for years.
Buying for the fleet you will have, not the one you have
The purchase you are making now sets the shape of a fleet you will be living with for several hiring cycles. Choosing a model family that still exists in volume next year, keeping profiles narrow, holding a buffer and recording serials from the first unit are all cheap while the fleet is small, and all painful to retrofit.
Most of what separates a team that handles its fiftieth hire calmly from one that scrambles is decisions like those, made early and dully. Anyone reading up on how to buy computers for your business for the first time tends to arrive looking for a specification and leave with a process instead, which is the right way round.
The first wave is never really about the hardware. It is the moment you decide whether the next ten waves are going to be a decision or a habit.
