Employee retention

Share of last year's team still here today. Of the people who were on the team a year ago, the share still here today, based on hire and departure dates on file.

The team is the most expensive investment the agency makes, and every departure walks out with client context and takes months of ramp to replace. The working floor is 75% of last year's team still here. Read the direction with team sentiment rather than alone: in a slow hiring market people stay for reasons that have nothing to do with wanting to. It sits on this board against floor 75%.

This accounts for the people who left, out of the 8 who were here at the start, one person in all.

Each person is counted once, against the team they were recorded in. Somebody hired and gone inside the same period is not here, because the share measures who stayed out of who was already here. By team, the heaviest loss was Account Management, which lost 1 of 3. Three from a team of four and three from a team of forty are the same number and different businesses, which is why the team size is beside it. Departures are counted against the people who were already here at the start of the period. Somebody hired and gone inside the same year is not in this cohort, which is what the share measures and not a judgment about the hire. No pay information travels with this list.

Like the client version, this reports on decisions already taken, and it reports them late. What sits behind it is rarely one thing anybody could have pointed at on the day: workload that stayed high past the point it was supposed to be temporary, a stretch of work nobody found interesting, or a manager. The pattern in WHO left is what separates those, and it is usually obvious the moment the departures are named: one team, one role level, one manager. On this workspace the departures were Maren Voss, and by team the heaviest loss was Account Management, which lost 1 of 3. The team survey is not on this board, so the comparison this calls for cannot be made here. Read this next to the team survey when there is one, because a share that holds up while the survey does not is the pattern worth catching. People who feel stuck and people who feel secure produce the same reading here, and only one of them is good news.

Watch the live signals between readings and treat this as the scoreboard. Workload balance and the team survey both move well before anybody resigns.

The structural problem with people metrics is the lag. By the time the scoreboard moves the ramp cost is already committed, and a replacement takes months to reach the output of the person who left. The signals that lead it show up weekly, and the one that matters most in a services firm is how the work is spread, because the people who go first are usually the ones carrying the most and complaining the least. Acting on those is the change. This tile only tells you afterwards whether it worked.

This closes no measured driver above, and does not claim to.

Every departure walks out with client context that was never written down and takes months of ramp to replace, and the cost lands in two places that do not look connected to it. Income per person falls, because headcount refills long before output does. And client relationships thin, because the person who knew the account is gone and the handover was a document. The compounding version is the one to watch: whoever covers for a departure is the most likely next one, so a single loss on a stretched team is rarely a single loss.