Managing someone else's calendar without crossing a line

Someone managing another person's calendar makes dozens of small decisions a week, and almost none of them are difficult in themselves. The difficulty is that the boundary of their authority has usually never been stated, so every decision carries a quiet risk: accept something the owner would have refused, or check something they expected you to handle, and either way you have got it wrong in a way that only becomes visible afterwards.

This produces two failure patterns and both are costly. The cautious assistant checks everything, which makes them slower than no assistant at all, because every request now takes two conversations instead of one. The confident assistant decides everything, which works until they accept a meeting the owner had specific reasons to avoid, and the resulting conversation damages trust disproportionately. Neither pattern reflects poor judgement. Both reflect the absence of a stated policy, and the fix is a single conversation that most working relationships never have — not because it is difficult, but because the need for it only becomes obvious after something has already gone wrong.

The categories that can almost always be handled independently are the ones where the owner's preference is either irrelevant or already known. Accepting a recurring internal meeting they have always attended. Declining a request that conflicts with something already confirmed, since the conflict decides it. Moving an internal meeting between colleagues where nobody is inconvenienced. Answering a request for availability with actual availability. Booking travel and logistics around an already-agreed commitment.

What these share is that a reasonable person would make the same call, so the decision carries no real discretion. Explicit permission on these is what makes an assistant genuinely useful rather than a relay, and it is worth the owner enumerating them rather than assuming they are obvious. The distinction to hold onto is between decisions that follow from facts already established and decisions that require knowing what the owner wants. The first kind can be handed over completely and permanently. The second kind cannot be handed over by a general grant of trust, however much trust exists, because trust does not supply the information needed to make the call.

The requests that need checking are the ones where the answer depends on information the calendar does not contain. Anything involving a relationship you cannot see the state of — a client who is unhappy, a supplier mid-negotiation, someone the owner is deliberately keeping at a distance. Anything that commits the owner to travel, since that consumes far more than the meeting slot suggests. Anything from a person more senior than the owner, where a decline needs to come from them. Anything that displaces something already confirmed, because you cannot tell from the entry how expensive that displacement is.

The general rule that covers most cases: if the decision requires knowing why something is on the calendar rather than merely that it is, ask. A block labelled with a project name might be routine work or the final day before a deadline, and no amount of experience lets you distinguish those from the entry alone. This is also the reason the owner's side of this arrangement matters — a calendar where important commitments are labelled meaningfully makes far more of these decisions delegable, and one full of opaque entries forces questions that better labelling would have answered.

The most valuable thing a person managing a calendar does is protect the time that nobody sends an invitation for. Preparation is the clearest case: an hour-long client meeting with no preparation time before it is not an hour of commitment, it is an hour of meeting plus whatever preparation gets improvised badly at the last minute. Booking prep time before any significant meeting, as a standing rule rather than a case-by-case judgement, is usually the single highest-value habit in the role.

Travel time is the same problem in a more measurable form, and it is the most common cause of a technically valid calendar that cannot actually be executed. Two meetings across a city with nothing between them do not conflict on the grid and do conflict in reality. Lunch deserves protecting for a plainer reason: a person who has not eaten makes worse decisions in the afternoon, and lunch is the easiest block to sacrifice because it has no other attendee. Treating these as bookable commitments rather than as gaps is what separates a calendar that reflects a workable day from one that merely contains no overlaps.

Double-bookings happen, and the way they are handled matters more than preventing every one. The instinct is to resolve it quietly before the owner notices, which is usually the wrong move: choosing which of two commitments to break is exactly the kind of decision that depends on context you do not have.

A workable sequence is to establish which was confirmed first, since the earlier commitment has the stronger claim and this is often uncontroversial; then to check whether either can be moved without cost, which resolves a good proportion of cases outright; and then, if a genuine choice remains, to present it to the owner with the relevant facts rather than presenting a decision. 'Both of these are at eleven on Thursday, the client one was confirmed last week, the internal one has four attendees and could move to Friday — which would you rather keep?' takes fifteen seconds to answer and puts the judgement where it belongs. The part not to skip is telling whoever loses the slot promptly and directly, with an alternative attached. A double-booking handled with a fast, clear message is a minor administrative event; the same double-booking discovered by both parties arriving is a real embarrassment for the owner.

It is tempting to treat all of this as something that resolves itself as a working relationship matures, and to an extent it does — patterns get learned, preferences get absorbed. But there is a limit that experience cannot pass, and it is worth being clear about it. An assistant can learn everything observable about how someone works and still not know that this particular client is in a dispute nobody has mentioned, or that a meeting labelled routine is actually the one that matters this month.

That information is not discoverable from the calendar, however carefully it is read. It has to be communicated, which makes the useful conversation a short and recurring one rather than a single policy document: what is coming up that I should protect, and who should I be careful with at the moment. Thirty seconds at the start of the week removes most of the ambiguity that otherwise accumulates into a mistake. The general division holds — decisions that follow from established facts belong to the person managing the calendar, and decisions that turn on what the owner wants need the owner's input, or a standing instruction covering that specific case.