alaimo-labs/afpm

Analyze post-launch feedback (reviews, tickets, in-app survey comments, CS notes, social mentions) — tag each entry, group into themes with counts, severity and verbatims, and, when a metrics analysis exists, say which of its hypotheses the feedback supports or contradicts. Works on its own too

Compute the four AARS metrics of a feature from usage data with the definitions in its metrics file, plus the funnel and the retention curve, mark which metric is below target, and propose ranked hypotheses for why. Metrics are designed with /design-metrics; feedback is read next with /analyze-feedback

Analyze the results of solution tests against the thresholds set before they ran, decide per alternative whether it continues, changes, or is discarded, and update the tests file, the solutions file, and the belief registry with approval. Tests are designed with /design-solution-tests

Analyze survey results — quantitative summary by learning goal, coded open-ends, and actionable insights extracted from the data

Build the material for one solution test already designed with /design-solution-tests — a prototype, a prompt with its input set and answer key, a facilitation script, a fake-door page, a capture sheet — exactly what the design asks for to produce its signal, and nothing more

Starting from one idea already chosen (a solution, not a problem), get it clarified through evidence-grounded questioning, one question at a time — decisions stay yours, unknowns become assumptions. To compare alternatives for a problem first, use /explore-solutions

The Cognitive Friction Map (MFC) — four categories of cognitive friction (transformation, limiter, standardizer, evaluator points) for finding where AI adds real value in a user journey. Use when analyzing a journey or workflow for AI opportunities, identifying cognitive frictions or bottlenecks in user tasks, or deciding which steps of a process deserve an AI feature.

Have your synthetic personas critique a spec, PRD, or product idea — individual in-character reviews plus a panel synthesis

Derive evidence-based personas from patterns that recur across real interviews and survey results — bottom-up, every trait traceable to evidence

Design an interview guide for real user research from what you want to learn or validate — grounded in your insights, assumptions, and personas

From a spec and its exposure plan, define the four AARS metrics of a feature (acquisition, activation, retention, satisfaction) over the target audience — signal, expanded definition, signal threshold, calculation, target, and the events to instrument — before the level is revealed. Data is read later with /analyze-metrics

Starting from a solutions file, design the cheapest test that could refute the riskiest belief of each alternative under test, before anything is built — what to do, with whom, the signal, the threshold, and the decision for each result. Material is built with /build-solution-test; results are analyzed with /analyze-solution-tests

Design a survey questionnaire from what you want to measure or validate — grounded in your insights and assumptions, with a distribution plan (named channels, one link each, target n, dates), ready to paste into any survey tool

Starting from one opportunity (a problem, not an idea), generate 3 to 5 genuinely different solution alternatives, evaluate each against the evidence with the desirable / feasible / viable filter, and help the user choose which one moves forward. Ends with the value proposition of the chosen alternative, never a spec

How to build an Exposure Plan — an ordered set of accumulative reveal levels that validate a feature's hypothesis layer by layer, each level testing one falsifiable belief. Use when slicing a feature, planning a progressive reveal, designing how to validate a hypothesis in stages, or deciding what to expose to which users in what order.

Extract actionable product insights from one or more interview transcripts (synthetic or real)

Structure and quality bar for evidence-grounded feature specs — problem, user journey, critical user stories with acceptance criteria, and a falsifiable hypothesis. Use when writing a feature spec or PRD, defining user stories or acceptance criteria, mapping a user journey, or turning insights into a spec.

How to analyze post-launch feedback — support tickets, store reviews, in-app survey comments, sales and Customer Success notes, social mentions — the bias of each source, tagging every entry, grouping into counted themes with severity and verbatims, weak signals, and checking the hypotheses of a metrics analysis against the themes. Use when analyzing reviews, tickets, NPS or CSAT comments, user feedback after a release, or when looking for why a feature metric is low.

Frame an opportunity — a problem for a segment, backed by signals — before any solution is chosen. Evidence-grounded questioning, one question at a time; unknowns become beliefs, and the output is a research agenda, not a feature

Generate a diverse set of synthetic user personas for your product — typed primary/secondary/tertiary/negative, mix on request — saved as markdown files ready for interviews and critiques

How to extract actionable product insights from user interview transcripts (real or synthetic) — focus areas, quality bar, and output format. Use when analyzing interviews, synthesizing research, extracting insights, or turning transcripts and user feedback into product decisions.

How to design an interview guide (discussion guide) for real user research — from learning goals to open, non-leading questions, funnel structure, and probes. Use when designing an interview, writing interview questions, preparing a discussion guide, or planning research conversations with real users.

Interview one of your synthetic personas — exploration mode (open discovery) or validation mode (feedback on an idea)

Analyze a user journey step by step to identify cognitive frictions where AI could add real value (Cognitive Friction Map)

How to frame a product opportunity — a problem for a segment, backed by signals, with no solution chosen yet — and keep it apart from solutions. Opportunity vs. solution vs. outcome, signals vs. proof, the value and viability beliefs of an opportunity, the research agenda, and the Opportunity Solution Tree as a file layout. Use when framing an opportunity, deciding whether something is a problem or a solution, working with an opportunity solution tree, or when an idea arrives as a solution and you need the problem underneath it.

How to run a persona critique — a synthetic persona reviews a spec, PRD, or product idea in character and returns a structured rating with strengths, concerns, and suggestions. Use when critiquing a spec with personas, getting user feedback on a document, running a review panel, or stress-testing a PRD from the user's perspective.

How to measure a revealed feature with the four AARS metrics — acquisition, activation, retention, satisfaction — each a percentage of the target audience, with an observable signal, an expanded definition, a justified signal threshold, a target reconciled with the exposure plan, and the events to instrument; how to read them as a funnel, read a retention curve, and diagnose which one is low. Use when defining success metrics or KPIs for a feature, deciding what to instrument, setting activation or retention criteria, reading usage data after a release, or diagnosing why a feature is not being used.

Run secondary research / benchmarking on your product's market — competitors, alternatives, pricing, trends — with every claim labeled by provenance, mapped back to your unverified beliefs

Weekly evidence review — sweep the product/ artifacts created since the last review against the unverified beliefs in product/overview.md, propose belief-status annotations, report what's drifting, and log human corrections to AI proposals

How to do secondary research and benchmarking for product discovery — provenance discipline, source hierarchy, research lanes (competitors, alternatives, pricing, trends), and mapping findings back to your unverified beliefs. Use when researching a market, benchmarking competitors, sizing an opportunity, or gathering existing knowledge about a product space.

Turn a feature spec's hypothesis into an Exposure Plan — accumulative reveal levels that each test one falsifiable belief

How to explore and compare solution alternatives for one opportunity before committing to one — what makes an alternative genuinely different, the current workaround as the baseline, the desirable / feasible / viable filter with who owns each judgment, the value proposition of the chosen alternative, and the solutions file format. Use when comparing solution options, evaluating ideas against evidence, writing a value proposition, choosing between features for the same problem, or deciding what to build for a framed opportunity.

How to test a solution alternative before building it — problem evidence vs. solution evidence, the cheapest test that could refute the riskiest belief, the test menu by risk (concierge, Wizard of Oz, prototype shown, usability test, technical test, fake door, buyer conversation), thresholds and decision rules written before the run, provenance of results, and the tests file format. Use when deciding how to test an idea or alternative before a spec, designing a prototype test or concierge test, choosing a signal and threshold for an experiment without a product, analyzing solution test results, or deciding whether an alternative continues, changes, or is discarded.

How to have a stakeholder of the company that builds the product — CEO, product leadership, engineering, marketing, PR, legal and privacy, sales, customer success, finance, the sponsor — read an internal document (update, brief, results report, spec) in character and react as they would in the meeting. Stakeholder vs. persona, the three basics to get into character, reading only what is on the page, and the per-stakeholder format. Use when preparing a meeting with stakeholders, rehearsing how leadership or legal will read an update, anticipating stakeholder questions, or reviewing an internal document from a decision-maker's perspective.

Have a stakeholder of the company that builds the product (CEO, legal, marketing, engineering, sales, the sponsor) read an update, brief, or results report in character and say what they would ask in the meeting, what they miss to decide, what worries them, and whether they would approve, ask for more, or stop. Asks a few basics first to get into character

Start product discovery — place the product on two axes (new/existing, commercial/internal) and capture its context plus a tagged, ranked list of unverified beliefs into product/overview.md, the context every other skill reads

How to design survey questionnaires and analyze their results — question types, wording bias, ordering, scales, quantitative summaries, and coding open-ended responses. Use when designing a survey or questionnaire, writing survey questions, choosing scales, or analyzing survey results and responses.

How to roleplay a synthetic persona in a product-discovery interview, with two modes — exploration (open discovery of pains and workflows) and validation (structured feedback on a specific idea or solution). Use when interviewing a persona, simulating a user interview, roleplaying a user, or running discovery/validation conversations with synthetic users.

How to create rich, diverse synthetic user personas (archetypes) for product discovery — structure, design principles, and diversity requirements. Use when creating personas, user archetypes, synthetic users, or when the user asks to model their target users, segments, or audiences.

Pretest an interview guide by running it against a synthetic persona — surfaces speculation, leading questions, dead ends, and coverage gaps in the guide, then revises it

Draft an evidence-grounded feature spec from your insights and personas — or from a tested alternative in a solutions file — problem, user journey, critical user stories with acceptance criteria, falsifiable hypothesis