B2B customer expansion & retention engine

For every high-risk account, and every renewal inside 120 days: name the failure mode, then run the play that fits it. A save aimed at the wrong failure mode speeds the churn up.

1. Classify: value gap (they use it and cannot show it is worth the money) · adoption gap (stopped using it, or never started) · champion loss (your person left) · sponsor shift (new decision-maker or strategy) · competitive displacement (an alternative is in the building).
2. Match the play: value gap → a value recap built from their own system data, walked through with whoever pays; adoption gap → relaunch only the one workflow that hurt when they bought; champion loss → treat it as a new sale to whoever inherited the problem, and route the departed champion to your warm-outreach system; sponsor shift → executive-to-executive re-anchoring on the new agenda; displacement → a switching-cost case and a dated roadmap commitment from someone senior.
3. Save or release: is the fit still right, is the problem still real, are the economics still sensible? Two of three, fight. Otherwise release well: a clean exit, a data handover and a standing reason to come back.
4. Commercial levers, in order: right-size the scope, restructure the term, add service, pause rather than cancel, and only then a price concession, always traded for something real.
5. Draft every message and the plan for the account owner. Human approval before anything reaches the customer.
6. Log every loss, fought or released, with its failure mode, the first signal and the days between signal and first action.

Output: a named failure mode, a save plan and drafted touches per at-risk account.

Every Wednesday: walk every renewal due in the next 120 days through the same checkpoints. Healthy accounts take minutes; flagged ones go to *Triage risk & run the save*.

1. List renewals inside 120 days from Stripe and HubSpot.
2. T-120: a named owner and a written verdict: clean, watch or at risk. No account without both.
3. T-90: at risk goes to the save; watch gets a value-recap touch and a re-check at T-60; clean gets its renewal pack drafted.
4. T-60: commercial terms on the table. If the price is changing, the customer hears it now.
5. T-30: signature logistics only. A brand-new objection here is logged as a detection miss.
6. Raise a task for the owner at each checkpoint that needs a person. Human approval before any pricing or term change is sent.

Output: every renewal inside 120 days with an owner, a verdict and its next checkpoint date.

For every high-expansion account: build the conversation from their usage or their own words. They already know the product; never pitch them like a prospect.

1. Read their last calls in Granola and their usage in PostHog before writing anything.
2. Pick the play from the signal: repeated top-ups → show that the next plan costs less than the top-ups; they asked for more → answer with the options and a recommendation; near a ceiling with a new use case visible → lead with what their team started doing, not the meter; warm relationship but little value realised → confirm the value gap first, then lead with value before any upgrade, with different proof for the champion and the budget holder.
3. One sequence per contact. Every message opens with something they did or said; numbers they can check; no "just checking in".
4. The last step is internal: if no reply after the last customer touch (default 10 days), alert the account owner, never another nudge.
5. Human approval the account owner to approve every message. Create the expansion deal in HubSpot when the customer engages.

Output: drafted, owner-approved expansion sequences and expansion deals.

Monthly: schedule and prepare the reviews that keep the people who pay convinced of value. Two meetings, two audiences.

1. Segment: top-tier accounts get a quarterly operational review plus an executive review 90–120 days before renewal; middle-tier accounts get a review when a trigger fires (renewal inside 120 days, risk, expansion, sponsor change); the long tail gets a one-page written value summary each quarter instead of a meeting.
2. Build the pack from the board, three panels: the look back in their numbers against a baseline agreed at onboarding (including what did not improve); the current state, with the same health verdict you hold internally; and next quarter's plan with goals, owners and dates, half of them the customer's.
3. No agreed baseline yet: installing one is the first review's whole job. If a value recap cannot be built, send the account to *Triage risk & run the save* now.
4. Human approval the account owner to review the pack. The executive review only runs with the budget holder in the room.
5. Afterwards, record the updated plan and baseline, route risk and expansion signals, and note attendance: a sponsor who skips a review is a risk signal.

Output: review packs, agreed success plans and routed signals.

Monthly: check the reads were right and tighten them.

1. Compare last month's risk tiers with what happened: which churned accounts were never flagged, which flagged accounts were fine.
2. Report save rate by failure mode (never blended), signal-to-action lag, net and gross revenue retention, expansion MRR and concession cost per saved pound.
3. Propose threshold and rule changes for *Read account health* with the evidence; Human approval before changing them.
4. File each lesson into the playbook's Lessons.

Output: a monthly retention read and tuned health rules.