Most work friendships are not really friendships. They are functional alliances with a warm finish — people you eat lunch with because the alternative is eating alone, people whose names you know and whose complaints you can anticipate. That is not nothing. It is civil, it reduces friction, it makes the hours easier. But it is also not the thing that calls you when your parent dies or asks the question your therapist has not gotten to yet. Most work relationships stay in that middle register, and that is fine. The ones that cross are a different matter.
The crossing, when it happens, tends to happen obliquely. It is rarely announced. Nobody sits down and says, "I think we should become actual friends." More often there is a moment — a project that runs long, a dinner after a bad day, a frank exchange that was too honest to take back, a shared laugh at something no one else in the office would understand — and the relationship quietly re-sorts itself. The organizational context is still there, but it has become furniture rather than structure. What was defined by shared employer becomes defined by shared recognition.
The particular difficulty with work friendships that become real is the institutional overlay. You found each other inside a structure that assigned you roles, incentivized certain behaviors, and created a common enemy called "the organization" that could be complained about together. Some of that shared complaining did real bonding work — griping is a low-cost way to establish that you see the same absurdities, which is no small thing. But when the job ends, or changes, or when one of you moves up and the other does not, the common substrate shifts. The friendship then has to prove that it was about the person, not the position.
This is the test most work friendships fail. Not dramatically, with rupture, but quietly, with attrition. The invitations taper. The replies get slower. The shared references lose currency. You realize you have been talking about people and situations your friend no longer knows, and they have been doing the same. The overlap that felt bottomless when you were in the building together turns out to have been topical. When the topic ends, so does the friendship.
The ones that survive this test tend to have made at least one prior investment outside the work context — dinner where the office was not the subject, a trip, a crisis assisted, a secret shared that had nothing to do with the job. These investments are what build the second floor. The first floor, the professional one, is load-bearing as long as you are in the same building. The second floor holds after you leave.
There is also something specifically valuable about a friend who knew you in a professional context. They saw you perform, and they liked you anyway. They know whether you are competent and whether you are kind under pressure, because they watched you under pressure. The version of yourself you present at work — more guarded, more strategic, more aware of being evaluated — is a real version. A friend who liked that version and then found the others is a friend who has seen a fuller picture than most. There is a specific kind of trust that comes from being known in conditions that are not designed to be flattering.
What tends to surprise people is how sustaining these friendships become in middle age, when the social circumstances that produce new friendships — school, shared housing, proximity — have mostly closed. Work is one of the few remaining places where adults are reliably thrown together with people they did not choose. When those collisions produce something real, it is worth naming and protecting. The work-friend-who-became-real is often the last friendship you made in a period when making friendships was still structurally possible. For many adults, they are not replaced.