Departing Clients Are Your Sharpest Auditors — If You Ask Before the Door Closes

When a client gives notice, the machinery of a professional firm moves quickly. There is a partner huddle, a file review, a "lessons learned" page added to a pack that will be read once. What rarely happens is the obvious thing: a properly structured conversation with the client themselves.

That is an odd omission. A departing client is the only person who has observed your firm from beginning to end, at close quarters, and no longer has any commercial reason to be diplomatic. Most firms treat that asset as a filing matter.

The post-mortem is the wrong instrument

An internal review answers the question "what did we get wrong?" It does so from inside the organisation, using the documents the firm chose to keep. Notices of termination, however, are rarely about the documents. They are about an accumulation of small experiences — a partner who stopped attending, a report that arrived three days late for six consecutive months, a junior sent to a meeting the client believed was a senior conversation.

None of that surfaces in a file review. All of it surfaces in forty-five minutes with the person who paid for it.

Four ways exit conversations fail

Too late. By the time the review is scheduled, the client has appointed a replacement, reframed the decision as obviously correct, and lost interest in explaining it. Memory is not neutral; it reorders itself in favour of the choice already made.

The wrong person asking. The relationship partner cannot credibly conduct this conversation. They are a party to it, and the client will not tell them the unvarnished version to their face. Politeness will be mistaken for satisfaction.

The wrong framing. Nothing collapses candour faster than a research call that turns into a pitch at minute twenty-five. The client relaxes, you reach for the deck, and the rest of the conversation is theatre.

The wrong method. A short online form with a rating scale is not an exit interview. It is a survey, and it will tell you that the client rated you three out of five, which you already suspected.

Who should sit in the room

One person, occasionally two, and never the account lead. The right profile is senior enough to be credible, distant enough to be honest, and without a commercial stake in the answer. A partner from an unrelated service line, a managing director, or a non-executive director all work. For the firm's largest accounts, an independent third party is worth the modest cost: clients say things to a neutral reviewer that they will not say to a competitor's partner.

Two is the maximum. Beyond that it becomes a panel, and panels produce testimony rather than conversation.

Earning candour

State the terms at the outset and mean them. This is not a sales call. We will not pitch. We will not argue with anything you say. We will send you a written summary of what we heard and what we intend to change.

Timing matters. Four to six weeks after notice is usually the sweet spot: recent enough to be accurate, distant enough for the decision to feel settled. Ask for a call rather than a form, and for forty-five minutes rather than fifteen. Five questions will yield more than twenty-five.

There is a small courtesy that pays disproportionately: offer the client something of use in return. A short debrief they can put in front of their own procurement committee, or a summary of how their peers are structuring similar arrangements. It reframes the conversation as an exchange rather than an autopsy.

Five questions worth more than a satisfaction score

1. When did you first sense this might not work? The date is often more revealing than the reason. A fissure that opened in month two and closed in month fourteen tells you something about your escalation routes.
2. What did you tell yourselves before you told us? This surfaces the internal advocates who defended you and the moment they stopped.
3. Who else did you speak to, and what did they say that we did not? Competitor claims are useful intelligence, whether or not they were true.
4. What would have had to be true for you to stay? The hypothetical removes blame and produces a specification.
5. If you were advising your successor, what would you warn them about working with us? Blunt, memorable, and almost always answered.

From anecdote to pattern

A single exit interview is a story. Ten is a dataset. The discipline lies in the log: reason code, the stage at which the relationship first fissured, service line, lead partner, fee band, tenure. Reviewed quarterly by the leadership team — not filed by marketing — those fields begin to show where the firm is losing work it should have won or should never have pursued.

Patterns tend to appear in three places. In qualification, where firms chase work outside their genuine capability because a partner was persuasive in the meeting. In staffing, where the client believed they were buying senior judgement and received leverage instead. In service design, where reporting cadence and responsiveness were never designed at all but simply inherited.

Doing something with it

Intelligence that changes nothing is worse than no intelligence, because it teaches your people that the exercise is ceremonial. Pick one change per quarter, make it visible, and tell clients you made it. Feed the objection language into proposals, so that the next pitch answers the question the last client actually asked. Revisit the onboarding sequence, since most fissures open in the first ninety days and are never formally closed.

And resist the temptation to convert the whole exercise into a win-back campaign. If you ask for candour and then pitch, you have taught the market that your research is a sales tactic, and you will not be given candour again. If the client raises a return, fine. Let them raise it.

A modest governance metric

One number is worth adding to the board pack: the count of structured exit interviews completed as a proportion of notified losses. Not a target, not a scorecard, just a figure that gets read aloud. Firms that measure it conduct them; firms that do not, write post-mortems.

The client who leaves holds information no internal review will ever produce. Handle the conversation properly, and a lost account becomes a piece of governance.