Keep & Grow Customers

Six sequence types anchored to your real activation milestone — welcome, activation, adoption, stall-recovery, expansion, milestones. Activation-first, helpful, never oversells your product.

Deflect the repetitive tickets, escalate the rest, always offer a person. An honest escalation rule + a deflection playbook, AI-ready knowledge-base templates, a deflection & CSAT tracker, and a prompt pack. No-code, no bot that traps customers, no guaranteed rate.

Exit interview + classifier + retention analyzer + 5 win-back sequences by churn reason. 48 hours to clarity.

Visa cut the merchant Excessive VAMP ratio from 2.2% to 1.5% on April 1, 2026 — comfortable in March, in violation in April on the same volume. This gate computes your Visa VAMP and Mastercard ECP ratios from your own numbers against the thresholds you confirm and reads each merchant account HEADROOM / APPROACHING / OVER THE LINE (worst network wins, Visa's 1,500-event monitoring floor respected). Prevention is scored separately — six weighted controls roll up HARDENED / SOFT / EXPOSED and name the fix-first move, but never average into the verdict, because winning a representment does not remove a counted chargeback. The shipped sample proves it: prevention 100 / HARDENED, ratio OVER THE LINE. Thresholds are buyer-confirmed editable cells so it stays right when the line moves again. Zero-dependency Python engine + workbook + ratio-math SOP + monthly-run runbook + a 3-MID sample. Grades an account's numbers, never a person; reads no live processor data, files and sends nothing. Not legal or payment-compliance advice.

The MCP-driven workflow that revives a dormant HubSpot or GoHighLevel list. Four Claude Skills read your contacts, rank every one (Revive now / Nurture / Let go / Do not contact), design an honest segment-matched offer, and draft the re-engagement sequence in your voice — with a suppression gate that holds back anyone opted-out or unreachable, whatever their value. It ranks and drafts only; you clear consent and send. Pairs with the Reactivation Readiness Engine (consent gate). Not legal advice.

The MCP-driven workflow that clears every new client for takeoff. Four Claude Skills read your signed deal from HubSpot or GoHighLevel, score whether the engagement is truly ready to kick off (Go / Almost / Not ready / Hold kickoff), and draft the whole onboarding packet into your own Google Drive — with a blocker gate that holds the kickoff until the contract, deposit, and scope are locked. It drafts and stages unshared; you share and send. Not legal advice.

Turn happy customers into a predictable, lowest-CAC referral channel, compliantly. The advocacy loop, a referral-economics model (referral CAC, net value, viral coefficient), reward + disclosure templates, and GHL wiring. Rewards honest referrals, never positive reviews. Not legal advice.

Is your advisory firm’s referral engine systematic or running on luck? A deterministic diagnostic — the grade-it companion to the build-it Advocacy Engine. Mark six dimensions 0/1/2 on evidence — ask cadence & triggers (18, growth), centers-of-influence pipeline (20, growth), advocacy identification (16), tracking & attribution (14), compliance & disclosure (20, gate), reciprocity & nurture loop (12) — for a 0-100 score banded SYSTEMATIC / LEAKY / AD HOC. Two worsen-only overrides catch what a score hides: a compliance-zero gate (an indefensible compensated arrangement forces AD HOC) and a no-engine floor (no ask AND no COI pipeline forces AD HOC). Portfolio rolls up to the worst firm (ALL SYSTEMATIC / GAPS TO CLOSE / RUNNING ON LUCK) with the fix-first dimension. Runnable Python engine + workbook + assessment & build-up playbooks + a 5-firm sample. Grades a referral system, never people, and rules on no specific arrangement. Marketing-Rule-aware, not legal advice.

Grow net revenue retention by expanding the customers who are succeeding: a health-gated expansion engine, an NRR/GRR calculator with a band verdict, expansion plays, value-first scripts, and no-dark-patterns guardrails. At-risk accounts get a save first. Not legal advice.

The compliance gate that runs before you wake a dormant list. Scores every contact CONTACT / REPERMISSION FIRST / SUPPRESS and gates the whole list GO / FIX / HOLD — from each contact's own consent and risk fields, not a guess. Ships a runnable Python tool, a workbook, two playbooks, and a sample list. It's the gate, not the sender; reduces TCPA/CTIA risk (not legal advice).

The pre-sale decision layer for captured leads. A runnable Python engine + a workbook score each lead 0–100 on six weighted levers (ICP fit, need signal, engagement recency, engagement depth, authority, timing) and return ADVANCE / KEEP NURTURING / RECYCLE / DROP — with a stale-engagement gate that holds a high-fit but cold lead in nurture (fit is not readiness). Routes the rest to the right nurture track. Reads and scores only; it never sends or edits your CRM, and doesn't grant permission to contact — confirm consent first. Your marks, no baked-in benchmark. Not legal advice.

Six Claude skills spanning support and success. Triage, on-brand reply drafter (never invents), macros & KB, reply QA, churn-risk flagger, customer health check.

One master prompt + five focused prompts that turn any meeting transcript into summary, decisions, action items, follow-up emails, and a standup-ready roll-up. Honest by design.

Codify your voice into rules a machine can check. Voice Extractor + 8-section Voice Spec + On-Brand Writer + 100-point QA Checker + Voice Transfer. Worked example: 22 -> 92.

Four tools for the whole retention lifecycle — onboarding sequences, support & success skills, churn autopsy kit, brand voice engine. One honest standard across every customer touch.