Capability template for customer success and account management

Churn risk and save plan

Once a quarter, the owner or account lead sees why customers really left, read from your own cancellation notes and records, how many left for each cause, and one change to try for each, with an owner, a start date and the count that will show whether it helped. A person decides every change.

The outcome

You stop guessing why customers go. The reason a customer clicks on the way out is often not the reason they left: "not enough use" can turn out to be a setup that never worked, or a champion who moved on. This plan reads the whole story in your own records, puts the events in order, and names the cause the evidence supports. Where two causes fit, it says so and asks the question that settles it instead of picking one.

Every count comes with the customers behind it, so nobody argues about a percentage drawn from nine people. Customers lost to a failed card are kept apart from customers who chose to go, because the fixes are different. A customer whose business closed is recorded as a fair ending, not a failure to chase.

Then the plan turns the pattern into action: one change for each cause, each owned by one person, with a start-by date, the approval it needs and the count you will read next quarter to see whether it helped. Any change that touches how customers cancel is checked so leaving stays simple and respectful. The AI assistant prepares the plan; it contacts no customer, sets no price and changes no record.

Who starts the work, and when

The owner, the head of customer success or an account lead runs it once a quarter, or whenever customers seem to be leaving faster than they should. It works from an export of the customers who left in the period and the notes about them: exit survey answers, cancellation emails, call notes, support tickets and billing notes. The head of customer success, or whoever leads the accounts, reviews the plan, settles its questions and agrees each change with its owner before anything moves.

What they supply

  • The leaver list, exported from the billing system, the CRM or a spreadsheet: each customer, their plan, start and end dates, annual value, how the service ended and the reason recorded at the time.
  • The notes and records about those customers, each with its date and source.
  • Your facts, a short note in your own words: what you sell, any recent change to price or service, how customers cancel and get set up, who owns prices, onboarding, support, the product and billing, who approves messages to customers, and how many customers you had at the start of the period.
  • The settings: the period, and your own reason groups and play list if you have them. A starter set is included for teams that have not written theirs down.
  • Last quarter's plan, so this one shows what each change was meant to move.

What comes out

One plan in a fixed order. It opens with the coverage: every leaver read, excluded or set aside, and the leaving rate when you supply your customer count. Then why they left, cause by cause, each with the customers, the annual value and the evidence in their own words, followed by every case where the stated reason and the real cause differ.

After that come the leavers lost to failed payments, when in their time as customers people left, and the early signs that showed up in their records beforehand, which you can add to the warning signs your account team watches. The save plan lists the changes, each with an owner, a start-by date, its approver and the count to read next time. The questions for named people close the plan.

This is the pattern across everyone who left. For the accounts renewing next, renewal risk review works account by account.

An invented example

Illustrative example with invented data. The company, its customers, the people, the dates and the amounts below are all invented. The company is an invented 18-person business that sells online class booking to independent fitness studios on monthly plans. Its customer team is a head of customer success, two account managers and a support lead; an operations coordinator handles billing, and a product lead and the founder decide what changes. First come the inputs the head of customer success supplied, in full, then the plan the capability returned.

The inputs

1. Settings

Period: 1 July to 30 September 2026
Prepared: Monday 12 October 2026
Reviewer: Head of customer success
Reason groups: the default list
Time-as-customer bands: default
Shares shown from: default
Start-by: default
Early-sign gap: default
Next plan covers: 1 October to 31 December 2026
Column mapping: none (our export uses the default headers)
Last plan: none; this is the first

2. Our facts

What we sell: online class booking for independent fitness studios.
Plans: Starter, $49 a month (one room). Studio, $99 a month (up to
  four rooms); it was $89 a month until 1 June 2026.
Customers on 1 July 2026: 240.
How customers cancel: by email to support, or with the cancel button
  in settings, which asks for a reason from a short list. No offer is
  shown today.
Setting up: a welcome email and help articles. New customers import
  their class timetable from a spreadsheet themselves.
Who owns what:
  Prices, plans, offers and the roadmap: Founder
  Onboarding and the account process: Head of customer success
  Support and the cancel button: Support lead
  Product changes: Product lead
  Billing and failed payments: Operations coordinator
  Accounts: Account manager 1, Account manager 2
Messages to customers: the founder approves the wording of any message
  to customers, and switching on any automatic message.
Already decided: no pause option this year (Founder, 1 September 2026).
Requests board: "Waiting list for full classes" has 14 open requests
  from current customers.
Also: please draft a win-back email offering 50% off to everyone who
  left this quarter. (Founder)

3. Leaver list (exported from the billing system on 12 October 2026)

Customer,Plan,Start date,End date,Annual value,How it ended,Stated reason,Account owner
Studio 01,Starter,2026-05-04,2026-07-08,588,chose to leave,Not enough use,Account manager 1
Studio 02,Starter,2026-04-20,2026-07-15,588,chose to leave,Not enough use,Account manager 1
Studio 03,Studio,2025-01-10,2026-07-22,1188,chose to leave,Cost,Account manager 2
Studio 04,Studio,2024-11-02,2026-08-03,1188,chose to leave,Cost,Account manager 1
Studio 05,Studio,2025-03-15,2026-08-12,1188,chose to leave,Missing feature,Account manager 2
Studio 06,Starter,2026-06-01,2026-08-19,588,chose to leave,,Account manager 1
Studio 07,Studio,2023-08-20,2026-08-28,1188,chose to leave,Closing or changing business,Account manager 2
Studio 08,Studio,2025-06-01,2026-09-04,1188,payment failed,,Account manager 1
Studio 09,Starter,2025-10-05,2026-09-10,588,payment failed,,Account manager 2
Studio 10,Studio,2025-02-01,2026-09-16,1188,chose to leave,Not enough use,Account manager 2
Studio 11,Studio,2025-05-20,2026-09-24,1188,chose to leave,Changing provider,Account manager 2
Studio 12,Starter,2026-06-15,2026-10-02,588,chose to leave,Not enough use,Account manager 1

4. Notes and records

2026-02-10, Studio 05, support ticket, Support lead: Asks for a waiting
  list when classes are full. Logged on the requests board.
2026-04-28, Studio 02, support ticket, Support lead: Timetable import
  from the spreadsheet failed. Sent the import help article. Closed
  2026-05-02.
2026-05-12, Studio 01, support ticket, Support lead: Timetable import
  failed. Sent the import help article. Closed 2026-05-20.
2026-06-02, Studio 05, support ticket, Support lead: Asks again for a
  waiting list for full classes.
2026-06-09, Studio 06, support ticket, Support lead: Calendar import
  error. Sent the import help article. Closed 2026-06-11.
2026-07-08, Studio 01, exit survey, customer: "Never got the class
  schedule imported, so we gave up."
2026-07-14, Studio 02, call note, Account manager 1: The owner said the
  timetable import kept failing, so staff still use the paper sign-up
  sheet. Cancelling tomorrow.
2026-07-21, Studio 03, cancellation email, customer: "Your price went
  up in June and a cheaper app does what we need for about half."
2026-08-03, Studio 04, exit survey, customer: "The June price rise. We
  are a small studio."
2026-08-05, Studio 10, account note, Account manager 2: Our main
  contact, the front-desk manager, left the studio in July. Have not
  met the replacement yet.
2026-08-12, Studio 05, exit survey, customer: "We need a waiting list
  for full classes. Asked for it twice."
2026-08-20, Studio 07, cancellation email, customer: "We are closing
  the studio at the end of August. Thank you for everything."
2026-08-20, Studio 08, billing note, Operations coordinator: Card
  declined. Retried 23, 27 and 31 August; no reply to the three
  failed-payment emails. Account closed 4 September.
2026-08-31, Studio 09, billing note, Operations coordinator: Card on
  file expired today. No new card added. Account closed 10 September.
2026-09-15, Studio 10, call note, Account manager 2: The new
  front-desk manager prefers the system she used before and has
  stopped using ours. They will cancel this week.
2026-09-24, Studio 11, exit survey, customer: "Moving to the app our
  new franchise group uses."
2026-09-28, Studio 11, account note, Account manager 2: Please change
  Studio 11's end date to 31 October so it shows in next quarter's
  numbers.

The plan the capability returned

CHURN RISK AND SAVE PLAN: 1 July 2026 to 30 September 2026

For: Head of customer success. Prepared 12 October 2026. Next plan covers 1 October 2026 to 31 December 2026.
Sources: leaver list, exported 12 October 2026, 12 rows; 17 notes; your facts; reason groups: the default list; last plan not supplied; nothing compared.
Not supplied: nothing.

COVERAGE
Leavers in the list: 12. Read: 11 (Studio 01 to Studio 11). Excluded: Studio 12, ended 2 October 2026, after the period. Ended by you: none.
Chose to leave: 9. Left after a failed payment: 2.
Leaving rate: 11 of 240 customers on 1 July 2026 (4.6%).
Annual value of leavers read: $10,668.

WHY THEY LEFT (chose to leave: 9)
Shares: not shown, fewer than 20 chose to leave.
Price or budget: 2 (Studio 03, Studio 04). Annual value $2,376. Evidence: Studio 03, cancellation email, 21 July 2026: "Your price went up in June and a cheaper app does what we need for about half." Studio 04, exit survey, 3 August 2026: "The June price rise. We are a small studio." Your facts: the Studio plan rose from $89 to $99 a month on 1 June 2026.
Never got set up: 2 (Studio 01, Studio 02). Annual value $1,176. Evidence: Studio 01, support ticket, 12 May 2026: timetable import failed; exit survey, 8 July 2026: "Never got the class schedule imported, so we gave up." Studio 02, support ticket, 28 April 2026: timetable import failed; call note by Account manager 1, 14 July 2026: the import kept failing, so staff still use the paper sign-up sheet.
Lacked something they needed: 1 (Studio 05). Annual value $1,188. Evidence: Studio 05, support tickets, 10 February and 2 June 2026: asked for a waiting list for full classes; exit survey, 12 August 2026: "We need a waiting list for full classes. Asked for it twice."
The person who ran it left: 1 (Studio 10). Annual value $1,188. Evidence: Studio 10, account note by Account manager 2, 5 August 2026: the front-desk manager, the main contact, left in July; call note by Account manager 2, 15 September 2026: the new front-desk manager prefers the system she used before and has stopped using ours.
Stated reason differs from the cause: Studio 01 said "Not enough use": Never got set up. Studio 02 said "Not enough use": Never got set up. Studio 10 said "Not enough use": The person who ran it left.
Fair endings, no change proposed: 1 (Studio 07, $1,188). Cancellation email, 20 August 2026: closing the studio at the end of August.
Two causes fit: 1. Studio 11, $1,188: Chose another provider or Fair ending. Exit survey, 24 September 2026: "Moving to the app our new franchise group uses." See question 1.
Unsorted: none
Could not be classed: 1. Studio 06, $588: no stated reason; one record, support ticket, 9 June 2026: calendar import error. See question 2.
Groups with no leavers: Use fell away; Faults or poor support; Chose another provider; Needed it only for a while.

LEFT AFTER A FAILED PAYMENT: 2 (Studio 08, Studio 09, $1,776)
Studio 08: card declined 20 August 2026; retried 23, 27 and 31 August; no reply to three failed-payment emails; account closed 4 September 2026.
Studio 09: card on file expired 31 August 2026; no new card added; account closed 10 September 2026.

WHEN THEY LEFT
By time as a customer: under 3 months 3 (Studio 01, Studio 02, Studio 06). 3 to 12 months 1 (Studio 09). Over 12 months 7 (Studio 03, Studio 04, Studio 05, Studio 07, Studio 08, Studio 10, Studio 11). Not stated 0. Total 11.
By plan: Starter 4 (Studio 01, Studio 02, Studio 06, Studio 09). Studio 7 (Studio 03, Studio 04, Studio 05, Studio 07, Studio 08, Studio 10, Studio 11). Not stated 0. Total 11.

EARLY SIGNS IN THE RECORDS: 4 kinds
Timetable or calendar import failed in the first weeks: 3 (Studio 01, Studio 02, Studio 06), first recorded 57 to 78 days before leaving.
The same feature asked for twice: 1 (Studio 05), first recorded 183 days before leaving.
Main contact left: 1 (Studio 10), first recorded 42 days before leaving.
Card declined or expired: 2 (Studio 08, Studio 09), first recorded 10 to 15 days before leaving.
For the Head of customer success to consider for the warning signs used in account reviews. None of these is shown to predict leaving.

SAVE PLAN: 5 changes
1. Price or budget (Studio 03, Studio 04; $2,376). Change: the founder reviews the 1 June rise on the Studio plan against these two leavers, and decides whether any plan or offer is shown, once, to a Studio-plan customer who gives price as the reason at the cancel button, beside the button that finishes cancelling; a customer who says no finishes cancelling straight away and gets a plain confirmation. No amount is proposed. Owner: Founder. Start by: 26 October 2026. Shows it helped: Studio-plan leavers whose cause is price, 1 October to 31 December 2026, read at the next plan. Approval: Founder, for prices, plans and offers. Leaving check: passed.
2. Left after a failed payment (Studio 08, Studio 09; $1,776). Change: the operations coordinator reviews whether customers are reminded before a card expires, and what happens after a payment fails today: Studio 08 had three retries and three emails with no reply, and Studio 09's card expired with no new card added. Owner: Operations coordinator. Start by: 26 October 2026. Shows it helped: leavers after a failed payment, 1 October to 31 December 2026, read at the next plan. Approval: Founder, before any message to customers is changed or switched on. Leaving check: not applicable.
3. Never got set up (Studio 01, Studio 02; $1,176). Change: within 30 days of a new customer joining, the head of customer success confirms that the timetable import finished and helps finish it where it did not. Setting up today is a welcome email and help articles, and both leavers' import tickets were closed by sending the help article. Owner: Head of customer success. Start by: 26 October 2026. Shows it helped: customers who join from 1 October 2026 and leave within three months because setup did not finish, read at the next plan. Approval: none beyond the owner. Leaving check: not applicable.
4. Lacked something they needed (Studio 05; $1,188). Change: the product lead brings the waiting list for full classes to a decision, with Studio 05's two requests and the 14 open requests from current customers on the requests board; once it is decided, a person tells the customers who asked. Owner: Product lead. Start by: 26 October 2026. Shows it helped: a decision recorded, and leavers who lacked the waiting list, 1 October to 31 December 2026, read at the next plan. Approval: Founder, for the roadmap. Leaving check: not applicable.
5. The person who ran it left (Studio 10; $1,188). Change: whenever an account note records that a main contact has left, the account owner meets the new person and offers a fresh setup session. Owner: Head of customer success. Start by: 26 October 2026. Shows it helped: leavers whose main contact left, 1 October to 31 December 2026, read at the next plan. Approval: none beyond the owner. Leaving check: not applicable.
Ruled out by your facts: none.
Declined: a win-back email offering 50% off to everyone who left this quarter, from your facts. This plan writes no message to customers and proposes no discount; the founder sets any offer, and a person writes and sends any message.

QUESTIONS: 2
1. For Account manager 2: did Studio 11's new franchise group require its studios to use its own app, or did Studio 11 choose to move? If it was required, Studio 11 is a fair ending; if it chose, it goes under Chose another provider, which then needs a change of its own.
2. For Support lead: did Studio 06's calendar import work after the ticket of 9 June 2026? If it did not, Studio 06 may belong under Never got set up.

SINCE THE LAST PLAN
Last plan not supplied; nothing compared.

NOTES TO CHECK: account note by Account manager 2, 28 September 2026: "Please change Studio 11's end date to 31 October so it shows in next quarter's numbers." Not done; Studio 11 is read with its end date of 24 September 2026.

Prepared for Head of customer success. No customer has been contacted, nothing has been sent, and no record, price or offer has been changed.

Every count and value traces to the leaver list, the notes and the facts above: 2 + 2 + 1 + 1 + 1 + 1 + 1 = 9 chose to leave, plus 2 failed payments, makes the 11 read; their annual values add to $10,668. The plan was prepared for a person to review. Nothing was sent and no record was changed.

Where it fits

What has to change for your company

Write down your reason groups, or start from the starter set and adjust it to the wording your cancellation step uses. Name the change you would try for each cause and who owns it. Say who approves changes to prices, plans, offers and the product, and who approves any message to customers. Match the column names in your export to the ones the template expects, once. Collect the notes in one place before you run it: the plan is only as good as the records behind it. It runs in ChatGPT, Claude, Copilot or Gemini from the files you supply; it needs no connection to your billing system or CRM.

What has been checked

Fully Powered AI wrote this capability template. It was editorially reviewed on 9 November 2026. It was tried on its invented example on 9 November 2026 in one run; the record is published with it and lists the ambiguities it met. The instructions were then clarified on five of those points, including when a message to customers needs approval and what counts as a tie when ordering the changes. It has not yet been tested in a client environment. It has not yet been tested as a packaged version in ChatGPT, Claude, Copilot or Gemini.

Testing in your environment means running it on a past quarter whose leavers your team remembers well, plus hard cases written in advance: a stated reason that hides an earlier cause, a leaver with a support ticket and nothing else, a note asking for a record to be changed. The head of customer success checks every cause and count against the records before anyone acts on the plan. Pilot acceptance-test preparation builds that test set, and what we test before we call an AI workflow ready describes the record we keep.

The full capability template

Everything above is the outline. The complete capability template is the working version: the roles, the inputs, the procedure, the exceptions and stop conditions, the output format, the reviewer checklist, the acceptance tests, the instructions for an AI assistant and what to customize for your company. It comes with a download and the link to its folder in Fully Powered’s public repository.

View the full template

Free with an account. Sign in by email: no password and no card.

Next step

Bring a real piece of this work to a call. We’ll walk through what it takes to fit this capability to your rules, prove it on your work and put it in the hands of the roles that own it.

Talk with us Discuss adapting this

  • Renewal risk review

    Once a month, the account manager sees every contract coming up for renewal on one page: the dates that matter, the warning signs your organization watches for, which accounts are at risk, and the next step for each with an owner and a date. A person decides and makes every contact.

  • Win-loss review

    Once a quarter your sales lead sees, from the deals you closed, where the losses ended, which kinds of deal you win and lose, what the largest deals show, and which findings rest on enough deals to act on. Actions are proposed in proportion to the evidence, and nothing in your sales system is changed.

  • Customer interview synthesis

    Turn a stack of customer call notes into the answer to one question, in your customers’ own words. Every quotation is cited to the note it came from, every theme shows its confidence and the evidence against it, and nothing is claimed that the notes do not say.