product

Plugin product from marketplace greenblacked-ai in greenblacked/AI. Commands read from the repo's manifest — 2 days ago.

Skills (22)

• activation-analysis

  Define activation from your own data and find where new users stall: cohorts by sign-up week, candidate early actions scored against week-8 retention, the winning event and threshold proven with a randomised nudge test, not declared from a correlation, each drop-off checked for tracking artefacts fi

• churn-analysis

  Explain why customers cancel or shrink: split logo churn from revenue churn and voluntary from involuntary before anything else, cohort by acquisition month and segment, code reasons from cancellation notes, tickets and the prior 90 days of usage rather than the cancellation dropdown, and back-test

• competitor-signal-watch

  Infer where a competitor is heading from the traces it leaves in public: job postings counted by team, changelog and pricing pages archived and diffed over time, review-site themes, partner and integration pages. Every inference carries its source, the date it was captured and a confidence level; th

• concierge-mvp

  Deliver a product's outcome by hand to the first five to ten customers before automating anything, and let what happens decide what gets built. Log every manual step with the time it took and the customer's reaction, so the log becomes the spec; automate only a step that has run ten or more times an

• contact-reduction-review

  Treat support tickets as a defect list and cut the contacts at their cause: tag a month of tickets by contact driver, the customer's reason for getting in touch, capped at about twenty tags; cost each driver as volume times handle time; give each of the top ten one fix: product change, copy change,

• design-partner-program

  Run a small cohort of design partners or a customer advisory board that shapes the product without capturing it. Choose five to eight partners for fit to one segment, not logo size; write the give-get (a weekly 30-minute session, real data, a reference at the end, for early access and visible influe

• discount-discipline

  Put deal-desk rules around discounting so margin stops leaking one deal at a time: audit the last 50 closed deals for realised against list price by rep, segment and closeness to quarter end before writing any rule, then set a discount ladder with a named approver per level, a give-get rule so every

• dunning-recovery

  Recover failed subscription payments before they become churn: measure the involuntary share of churn first, retry by decline code (soft declines retried on a spaced schedule, hard declines sent to the customer), switch on a card account updater, send pre-expiry notices, run a three-email sequence o

• escalated-account-recovery

  Recover a key customer account that is angry and threatening to leave: one named owner who acknowledges within 48 hours, the complaint written back in the customer's own words and agreed before any fix is proposed, a written recovery plan with dated commitments and an executive sponsor on each side,

• fake-door-test

  Measure demand for an unbuilt feature or product with a fake door, smoke test or landing page, and decide from a rule written before launch. Set the pass threshold and sample size from what the build is worth; count only a costly action (a waitlist sign-up with a work email, a refundable deposit, a

• feature-sunset

  Retire a feature, plan, integration or whole product without stranding the customers who depend on it: pull usage by account before deciding, classify every account as unaffected, replaceable or blocked, set the end date from the slowest-migrating segment rather than engineering convenience, then ru

• jobs-timeline-interview

  Reconstruct why recent buyers or switchers chose your product, from first thought to first use, so positioning and roadmap rest on what happened rather than on opinions. Interview only people who bought or switched in the last 90 days; walk the timeline (first thought, passive looking, active lookin

• metric-anomaly-triage

  Decide whether a sudden move in a product metric is real before anyone explains it: a fixed elimination order of tracking or tag change, pipeline lag or late data, definition or filter change, bot or internal traffic, mix shift, and seasonality, each ruled out by one named query, and only then a rea

• metric-gaming-audit

  Stress-test product KPIs and key results for the cheap ways to hit them while harming customers: write the three cheapest moves that raise each metric while damaging the customer, pair each with a counter-metric and a data check that would expose it, require every key result to carry a baseline, wri

• poc-plan

  Scope and run a customer proof of concept or pilot as the vendor so it ends in a decision rather than drifting: success criteria taken from the buyer's own problem, measurable and signed by the buyer before any environment is set up; a fixed window of two to four weeks with a named customer sponsor

• price-change-rollout

  Raise prices or move existing customers onto a new price or plan without a churn spike: model revenue by cohort first (plan, tenure, size, discount, renewal date), choose grandfather-with-sunset or migrate-at-renewal per cohort, give notice of at least the renewal window (60 to 90 days for annual B2

• pricing-research

  Find out what customers will pay before a price is set: pick the method by sample size (under about 30 respondents, value interviews anchored on what the current alternative costs them; above it, Van Westendorp for the acceptable range plus Gabor-Granger for demand at each price), ask the buyer who

• request-triage

  Handle feature requests from sales and customers so real deal-blockers surface and every requester hears back. Record the job and the customer, never the feature wording; ask \"what will you do if we never build it, and would you sign without it?\" and count a blocker only with the customer's contin

• rfp-response

  Respond to an RFP, RFI, tender or customer security questionnaire as the vendor: score bid or no-bid within 24 hours before anyone writes a word, build the compliance matrix from every requirement before the narrative, answer from an approved-answer library where each answer has an owner and a revie

• roadmap-disruption-triage

  Decide what to do when a key customer, an executive or a competitor's move tries to force something onto a committed roadmap: write the trade-off ledger before answering (the item moving in, the named items slipping, their new dates, who must be told), sort the demand into promise, proof or pressure

• spec-premortem

  Attack a spec, PRD, launch plan or big bet before committing to it, then score the forecast afterwards. Assume it failed six months out; have each reviewer write reasons alone before any discussion; sort them into real risks, paper tigers and unspoken elephants; turn each real risk into a tripwire w

• win-loss-review

  Find out why deals are won, lost or stall, from the buyers rather than the CRM: separate no decision, often the biggest bucket, from lost to a competitor; sample every closed deal in the period rather than the ones reps pick; have someone outside the sales team interview buyers within 30 days of clo

Subagents (1)

• cancellation-note-reader

  Read exported cancellation notes, support tickets and success-team notes for a sample of churned accounts and return one primary code per account from the codebook the caller supplies, with the quoted line that decided it and logo and revenue totals per code, so the notes themselves never enter the

Similar plugins