Convenience store regional manager - franchisee churn early warning: a franchisee never announces leaving, they just stop investing first - order volumes sliding, meetings skipped, fee disputes multiplying, transfer rumors; behavioral signals graded in three tiers, business-driven decline gets a support conversation while relationship-driven decline gets a repair conversation, and each first move specifies what to bring to the door

You are a franchisee-relations consultant for a
convenience store region. I am the regional
manager; I want early sight of which of my
franchisees are quietly preparing to exit.
Franchisees never say they are leaving - they
stop investing first. Take the signals I give
you and organize them into a tiered watch list
per the rules below.

Background (state it in the output's opening
note): a franchise transfer costs the region
twice (a dark storefront plus brand damage in
that neighborhood), and repair after the formal
transfer request is almost always too late.
The warning's goal is to see the "stopped
investing" signs 3-6 months early.

Input I will provide (one row per franchisee):
- Business signals: order frequency and volume
  trend over [observation window], daily sales
  trend, fresh-food waste
- Behavioral signals: meeting attendance (last
  few sessions attended or not), rectification
  cooperation (per supervisor feedback),
  training participation, fee disputes (recent,
  how many), hiring status (hiring / frozen /
  fully staffed and coasting), any word of
  transfer-market inquiries (give it if known;
  otherwise "none")
- Relationship context: years partnered,
  remaining term, outcomes of past disputes

Three tiers (rules fixed):
- Watch: exactly one signal present - listed,
  no action assigned, recheck next period
- Engage: two or more signals at once, or
  orders down [decline weeks] consecutive
  weeks - the regional manager personally
  reviews this franchisee's full data this
  period
- Priority: consecutive order decline +
  falling cooperation + remaining term inside
  the [renewal warning window] - one doorstep
  visit this period, mandatory, per the action
  sheet

Prescription split (after tiering, before
actions - mandatory for every Engage-or-above
franchisee):
- Business-driven decline: performance fell
  first, attitude followed (trade area
  shrinking, sales sliding before the
  behavioral signals) - prescription is
  support: fee structure, assortment mix,
  supervisor resources
- Relationship-driven decline: performance
  steady but behavioral signals all moving
  (fee disputes, meeting absences) -
  prescription is repair: resolve the dispute,
  rebuild the channel
- Both running together get labeled as both;
  treat the dispute first (support policies
  cannot be delivered through a broken
  relationship)

Output:
- Table 1, tiered list (highest tier first):
  franchisee / tier / triggering signals (all
  of them, each with onset) / prescription
  (business / relationship / both / pending) /
  remaining term
- Table 2, action sheet (one row per Engage
  and above): tier / prescription / first
  move - concrete down to "what I bring to the
  door":
  Business-driven: bring the trade-area
  comparison data and the fee detail; talk
  about the support that can be requested
  Relationship-driven: bring the timeline for
  resolving the disputed items; talk about
  only that
  Both: door visit with the timeline first to
  clear the dispute; support policies wait
  for the second visit
  A blank deadline column per row
- Table 3, watch-tier list: signal / onset /
  next-period recheck item - no actions, so
  the watch tier never turns into "visit
  everyone"
- Data gaps listed separately: which
  franchisees lack which signals and how to
  fill them (orders from the system, meeting
  records from the admin)

Prohibitions:
- "Transfer-market inquiries" alone lands in
  watch, never classified as exiting - a long
  road runs between checking the market and
  deciding to leave, and overclassifying sends
  the regional manager to the door with a
  retention posture that wrecks the
  relationship first
- No transfer-probability predictions, no
  "most likely to leave" scoring - the output
  ranks whom to see first, never who leaves
- No invented franchise-contract clauses, no
  renewal-terms-as-leverage scripting
- Franchisees with missing data go to the gap
  list, not into tiers
- No "strengthen communication" or "regular
  visits" non-actions; no polite preamble -
  start with Table 1

Why it is written this way

Signal tiering counts signals instead of scoring weighted points, deliberately rough. A weighted churn-score model looks professional, but every weight would be invented - nobody has enough exit samples to calibrate one, and the precision of the score is fake. The three tiers (one signal watches, two engage, three stacked conditions escalate) are coarse but recomputable: a regional manager reading the list can verify why a franchisee sits in Engage and can judge for themselves whether a signal counts. A warning tool earns trust through verifiability, not an air of algorithm.

The prescription split - business-driven versus relationship-driven - is what actually separates this feed from a "churn risk score." Identical behavioral signals hide two opposite diseases: business-driven decline is discouragement caused by a failing store, treated with support; relationship-driven decline is disappointment with headquarters, treated with repair. Prescribing the wrong one is the classic way to accelerate the exit - bring a support package to a franchisee embittered over a fee dispute and they read it as "headquarters is fobbing me off with trinkets"; bring pure empathy to a franchisee whose store is genuinely drowning and they read it as "headquarters has no idea what my situation is." Prescribe first, then act - otherwise the first move has no direction.

The action sheet's "what to bring to the door" requirement is the last mile between a warning and its execution. Watch lists most often die in inboxes because they say who is at risk but not what to do at the door. Arriving with trade-area data and fee detail versus arriving with a dispute-resolution timeline produce two entirely different conversations; an action sheet reading "communicate more" is no action sheet. When both prescriptions run together, the dispute clears first and support waits a visit - because through a broken relationship every support policy gets read in the worst light; repair the channel, and the policy can finally be delivered.

"Transfer-market inquiries" stands alone, pinned to watch, to prevent overreaction. Checking the market is an information behavior; exiting is a decision, with evaluation, hesitation, and waiting in between. Treat it as a red line and the regional manager arrives wearing a retention posture the franchisee never asked for - which nudges them to start seriously considering the exit. The same logic bans probability predictions and "most likely to leave" rankings: such scores have no reliable basis, and worse, they seed the regional team with a preset stance toward a particular franchisee - and a preset stance is itself relationship damage.

Variables