Client health

The counterintuitive fact about agency churn is that performance is a lagging indicator of it. An account can hit every target for a year and be lost, because what decayed was the relationship rather than the results.

It works in the other direction too. Accounts miss targets for two quarters and renew, because the client trusted the explanation and could see the work. Results matter; they are not what the renewal decision turns on.

What moves first is operational and mostly invisible in a dashboard. Approvals sitting longer than they used to. Meetings rescheduled, then shortened, then delegated. A new name copied on emails with no introduction. Questions that used to come by phone arriving in writing. Replies getting shorter.

Each of those is small enough to ignore individually, which is why the mechanism has to be a routine rather than an instinct. Somebody has to be asked, every week, how each account feels, because the person delivering the work already knows and has nowhere to put it.

The instinct is to build a score: weight the inputs, produce a figure out of a hundred, track it over time. That tends to fail in a specific way.

A numeric score becomes an object of management in its own right. People tune the inputs, argue about weightings, and watch the number move rather than acting on what it represents. It also implies a precision that a judgement about a relationship does not have.

Three states work better: fine, needs attention, at risk. They are coarse enough that nobody argues about the boundary, they are hard to game because moving between them requires saying so out loud, and each one implies an obvious response.

The two fields that make it work are a named owner and a next action. Health without an owner is a status that belongs to everybody and therefore nobody, and without a next action it is a feeling recorded in a spreadsheet. Anything not green needs both, and a date.

A health field that nobody looks at is worse than not having one, because it creates the impression that somebody is watching. The value is entirely in the schedule, and weekly is the right frequency: often enough to catch a shift, short enough that it stays a few minutes rather than a meeting.

The format that survives is narrow. Anything not green, why, who owns it, what happens next, by when. Green accounts get named and skipped. That keeps the whole thing under ten minutes for most agencies and prevents it becoming a recital of concerns with no consequence.

The cultural condition is that raising a concern has to be safe and cheap. If flagging an account as at risk triggers an inquiry into whoever flagged it, everything will be green until it is gone. The status is information, not an accusation.

It is also worth reviewing the greens occasionally. An account that has been green for eighteen months and that nobody has spoken to at a senior level is not healthy, it is unobserved, and those are the ones that produce the email nobody saw coming.

A contact leaving is the highest-risk event in an agency relationship and is consistently under-reacted to, because on the surface nothing has changed: the contract, the work and the results are all the same.

What has changed is that the person who knew why the agency was hired, what had been tried, and which decisions were already made is gone. The replacement inherits an invoice and no context, and they usually arrive with their own view of how this should be done and sometimes their own agency.

Nothing about the existing relationship transfers automatically. The credibility built over two years belonged to a person, not to the account, and it has to be rebuilt from the beginning with somebody who is deciding in their first quarter what to keep.

The response is to treat it as a new relationship rather than a handover. A proper introduction, the history retold, the reasoning behind the current plan made explicit, and a senior person present. That is a few hours of work at the moment it is cheapest, against an account that is otherwise at real risk within two quarters.