agentleFS
agiea-brainagiea-brain/agency-skills/skills/bailey-darling/client-churn-autopsy/SKILL.md

client-churn-autopsy

Skill1 starsChanged 16 days agoMITView on GitHub

Use this skill in the week after a client leaves, while the record is still readable and before the story settles into whatever is least uncomfortable. Produces the real cause separated from the stated one, the point the engagement actually turned, what would have caught it, and whether this one is winnable back. Also use for "they churned", "why did we lose them", "post-mortem on the account", "can we get them back".

The file

We split this skill into the rules it gives an agent

10 rules. About 0 are common, 10 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈10

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read the engagement from the start before asking anyone what happened.

    line 19
  2. Distinctive:

    MUST read the engagement record before collecting anyone's recollection.

    line 48
  3. Distinctive:

    MUST record the client's stated reason verbatim and treat it as separate from the cause.

    line 49
  4. Distinctive:

    MUST identify a specific point where the engagement turned.

    line 50
  5. Distinctive:

    MUST distinguish underdelivery from invisible delivery from external change.

    line 51
  6. Distinctive:

    MUST state what signal existed in the record and when.

    line 52
  7. Distinctive:

    MUST give a win-back verdict with a condition attached.

    line 53
  8. Distinctive:

    MUST compare against prior losses for a pattern.

    line 54
  9. Distinctive:

    NEVER accept the stated reason as the finding.

    line 55
  10. Distinctive:

    NEVER conclude the client was difficult and stop there.

    line 56

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Applies after a client has given notice or gone quiet and not renewed. Produces the autopsy and a verdict on whether to pursue them later.

The play

  1. Read the engagement from the start before asking anyone what happened. The shared channel, the call notes, the reports that went out. Memory a week after a loss reorganises itself around whoever is least at fault, and the record does not.
  2. Write down the stated reason exactly as the client gave it. Budget, a reorg, bringing it in house, a change of priorities. Keep it, and treat it as a data point rather than the answer, because the stated reason is the one that is easiest for both sides to say out loud.
  3. Find the turn. There is almost always a specific week where the engagement stopped mattering to them: a champion left, a quarter went badly on their side, a deliverable landed late and the tone changed after it. Name the week and what happened in it.
  4. Separate the three causes that get conflated:
    • The firm underdelivered. Say so plainly, with what and when.
    • The firm delivered and nobody internally could see it. A different problem with a different fix, and the most common one.
    • Nothing was wrong with the work and their world changed. Real, and the version most likely to be winnable back later.
  5. Check what the firm knew and when. Was there a signal in the record that nobody acted on. Three check-ins with the same blocker, a report that went unanswered twice, a sponsor who stopped replying. The answer is usually yes, and that signal is the one worth wiring into how the firm runs.
  6. Name what would have caught it, specifically, and whether it is worth doing for every client or just for accounts that look like this one.
  7. Give the win-back verdict: not worth it, worth a check-in in six months, or worth a call now. If it is worth something later, write what has to be different before reaching out and the one sentence that would open it.
  8. Record the pattern against prior losses. One churn is an anecdote. Three with the same turn is the actual finding, and it is usually about how the firm sells rather than how it delivers.

What good looks like

  • First thing I check that most people skip: whether the champion was still there at the end. Half the losses I have looked at are not a work problem at all, they are a sponsor leaving and nobody noticing they had no second relationship in the account.
  • The mediocre version: writing down the reason the client gave, filing it under budget, and moving on. It costs nothing in the moment and it means the same loss happens again next year.
  • I know it is right when the autopsy names a specific week. If the cause is spread evenly across the whole engagement, the real cause has not been found yet.
  • Second test: somebody in the firm disagrees with it. An autopsy everyone nods at has usually landed on the comfortable answer.
  • Red flag: the conclusion is that the client was difficult. Sometimes true, almost never useful, and it ends the investigation exactly where it should start.

Rules

  • MUST read the engagement record before collecting anyone's recollection.
  • MUST record the client's stated reason verbatim and treat it as separate from the cause.
  • MUST identify a specific point where the engagement turned.
  • MUST distinguish underdelivery from invisible delivery from external change.
  • MUST state what signal existed in the record and when.
  • MUST give a win-back verdict with a condition attached.
  • MUST compare against prior losses for a pattern.
  • NEVER accept the stated reason as the finding.
  • NEVER conclude the client was difficult and stop there.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.