The first two weeks a remote employee starts to fall behind

A remote employee rarely announces that they are drowning, and the first signal is almost always a change in response time. Marek’s replies went from an hour to a day.

The week the replies slowed

Marek had handled the hardest customer escalations for two years and answered within the hour. In the first week of March his answers began arriving the next morning.

Nobody flagged it, because the work still got done. What changed was the shape of the answers: shorter, later, and with fewer questions attached.

Fewer questions is the part I noticed last. A person who stops asking has usually decided the asking costs more than the answer is worth.

Signals arrive before performance

Performance data lags by a month or a quarter. Behaviour moves in days, which is why the useful signals are cheap and easy to ignore.

Response time, message length, camera on or off, and the number of threads started. None of those appear in a quality report until much later.

We treat them as noise because they are not auditable. They are still the earliest evidence available, and the only evidence that arrives in time.

Check the calendar first

Before assuming a motivation problem, I counted his meetings. He had eleven recurring calls, four of them before eight in the morning his time.

He had also absorbed a colleague’s queue during a two-week absence and never mentioned it. The load had doubled and the schedule had not changed.

Almost every case I have handled started here. The person was not disengaging from the work, they were disengaging from a week that had no gaps in it.

The first conversation is not a warning

I asked for twenty minutes and put the purpose in the invitation. The sentence said I wanted to understand his week, not review his performance.

That sentence matters more than the meeting. Without it, a remote employee assumes a sudden call is the first stage of a formal process.

He arrived defensive and left forty minutes later having listed four things that were broken. Two of them I could fix the same afternoon.

Name the observation, not the person

I said I had noticed replies arriving a day later and asked what had changed. That is a description of events, and it invites an explanation.

The alternative is asking whether he is still committed, which turns an observation into a character question. It also ends the conversation on the spot.

Specific and recent beats general and historical. One week of timestamps is easier to discuss than a pattern I claim to have sensed since January.

Ask about load before commitment

My first question was what he had stopped doing to keep up. His answer was documentation, which explained a backlog I had blamed on someone else.

Only after that did we discuss whether the role still fitted. Asking about commitment first would have made the load question sound like an excuse.

The order is not politeness. It is diagnostic, because the answer to the load question determines whether the commitment question is even relevant.

The order of the questions

We settled on four questions and asked them in the same sequence every time. The sequence ran from what changed, through what he carried, to what he would stop.

Each answer narrows the next. By the fourth question we were discussing a specific queue rather than a general feeling of things being hard.

The sequence also protects the manager from arriving with a conclusion already formed. A conclusion makes every later question a request for agreement, not a question.

What I got wrong first

My first instinct was to schedule a daily check-in, which is the standard response and the wrong one. It added a meeting to a full week.

It also signalled that he was now being watched. That is precisely the condition that makes people hide problems more carefully and ask fewer questions.

The right move was removing two meetings and one queue. That took ten minutes and produced a visible change within four days.

Write the agreement down

We wrote five lines afterwards: what he would stop, what I would take, what would be reviewed, and on what date. He edited two of them.

A written agreement survives the week when both of us are busy. An oral one survives until the next urgent thing, which is usually Tuesday.

It also gives the next conversation a reference point. The discussion then starts from what changed rather than from two competing memories of March.

Set a two-week review

We booked a review for a fortnight later and named it a checkpoint. Nobody in that conversation used the word verdict, which was rather the point.

At that point three of the five lines had held and two had not. We changed the two, which is the normal outcome of a first attempt.

Two weeks is short enough to remember and long enough to see movement. A quarterly cycle would have buried the whole episode inside a rating.

When the answer is a different role

In a later case the answer was not load. The person had been promoted into coordination work and had stopped doing the work they were good at.

No amount of schedule repair fixes that. The honest conversation is about a role change, and it should happen before the performance record hardens.

We moved him back to delivery with a pay adjustment he accepted. He stayed two more years, which is longer than the alternative would have allowed.

The cost of waiting for the cycle

Marek’s case took eleven days from the first signal to a written agreement. The version where I waited for the review cycle would have taken eleven weeks.

In those eleven weeks a good person builds a story about a job that does not want them. That story is much harder to revise than a schedule.

The signals cost nothing to watch. The only expensive part is deciding, in the second week, that the observation deserves a conversation.

Quiet is the cheapest signal a remote team produces and the easiest to explain away. If someone went quiet last week, how many days before you could say what changed?