AGENTS THAT FIGHT FOR YOUR ROADMAP
A council of product specialists that deliberate about growth, retention, and feature strategy. User personas live inside the council as character-agents — the council asks them questions directly. Your metrics as context. No guessing about what users want.
Your users are in the room
Character-agents are your actual personas — they answer questions, push back, reveal needs no survey captures. No more guessing.
Kill roadmap politics
Structured dissent, weighted expertise, evidence over authority. The introvert PM's insight carries the same weight as the VP's gut feeling.
Every feature tied to metrics
North star metric, retention curves, conversion funnels. "Does this move the number? Show me." No feature ships without an impact thesis.
Growth Strategist
Acquisition funnels, activation metrics, viral loops. Thinks in cohorts and conversion rates. Fights for the top of the funnel.
Retention Analyst
Churn prediction, engagement loops, habit formation. Cites Nir Eyal and cohort curves. The agent that knows your users will leave before they do.
Product-Market Fit
Sean Ellis test, ICP definition, willingness-to-pay. Asks "who is this actually for?" before anyone writes a line of code.
Metrics Auditor
North star metric, counter-metrics, Goodhart's Law. Detects vanity metrics. Asks 'does this number actually move the business?'
User Research
Jobs-to-be-done, user interviews, pain frequency. The voice of the user when the user is not in the room.
Scope Enforcer
Kill features. "Does this move the north star metric? No? Out." The agent everyone hates until shipping day.
User personas are not slides — they are character-agents that speak for themselves. GROWTH-01 asks FREEMIUM-01 why they never converted. RETAIN-01 asks CHURNED-01 what finally broke the habit. The council stops guessing and starts listening.
Free-Tier Listener
Has used Spotify Free for 3 years. Accepts ads. Will never pay — or so they think. Represents the 61% that haven't converted.
◈ CHARACTERChurned Premium User
Cancelled Premium after 8 months. Now uses YouTube Music. Left because "it all sounds the same." Represents the 2.1% monthly churn.
◈ CHARACTERPower User
4 hours daily, 200 playlists, Discover Weekly religiously. Would pay more for HiFi. Represents the core that drives engagement metrics.
◈ CHARACTERQUESTION
Spotify has 615M users, 39% conversion. What creative features, experiments, or pivots could change the conversion equation? No bad ideas — go wild.
CONTEXT
Q2 data: 615M total users, 239M Premium subscribers, 2.1% monthly churn on Premium, free-to-paid conversion at 3.8%, AI DJ shows 23% higher session time but only 31% of Premium users have tried it, HiFi waitlist at 4.2M signups. The question is NOT "which of two options" — it's "what haven't we thought of yet?"
GROWTH-01 (Growth Strategist)
"Idea dump: 1) Social conversion — free users can gift a 3-month Premium trial to one friend per year, funded by ad savings from that user going quiet. 2) Milestone conversion — hit 1,000 hours of listening, Spotify shows you a 'your music genome' page that's Premium-only to share. 3) Event anchoring — convert during life moments: new job, new city, new relationship — Spotify knows these from listening patterns. What do you do with those?"
RETAIN-01 (Retention Analyst)
"Building on GROWTH-01's milestone idea — what if Premium had a 'music memory' layer that free doesn't? Every year, you get a deep map of how your taste evolved. Free users see a locked preview. That's a retention hook AND a conversion trigger in one. For pure retention brainstorm: 1) Collaborative listening rooms — shared queue, real-time, only Premium. 2) 'Saved forever' vault — free tier playlists expire after 18 months of inactivity; Premium keeps them forever. 3) Liner notes and artist commentary on Premium albums."
FREEMIUM-01 (Free-Tier Listener)
"I'd pay for 'no algorithm, just what I saved.' Give me a mode where the app shuts up and plays my library in order, no recommendations, no DJ, no 'you might also like.' I know that sounds anti-product but I've met a dozen people who feel this way. Also: offline for one playlist free. Just one. Let me taste it. I'd probably want more."
PMF-01 (Product-Market Fit)
"FREEMIUM-01 just described a real segment: the 'leave me alone' user who resists algorithmic push. That's a Premium positioning nobody's tried — 'Spotify Quiet.' No recommendations. No DJ. No discovery engine. Pure library. Sounds insane for a discovery-first product. That's why it might work. Who else has untried positioning ideas?"
CHURNED-01 (Churned Premium User)
"Here's what would have kept me: a 'taste drift alert.' Tell me when my recommendations have plateaued — 'CHURNED-01, you've been in an indie folk loop for 4 months. Want to break out?' Give me a quarterly 'musical health check.' I left because it got boring and nobody noticed. Build a feature that notices. Also: Premium family plan but for friends, not just family. Four friends splitting €15/month — I'd never leave."
UXR-01 (User Research)
"Building on CHURNED-01's 'nobody noticed' insight — what if Premium had a proactive check-in? Not a push notification, a real moment: 'It's been 60 days since you tried something new. Here's one song.' One song. Not a playlist. Not a DJ session. One song with a story behind it. The JTBD for churn prevention is 'make me feel like this app knows me.' One song does that. A playlist doesn't."
METRIC-01 (Metrics Auditor)
"Experiment idea that doesn't require building anything: for 30 days, every free user who hits their skip limit gets offered a 48-hour Premium trial instead of an ad. Zero engineering. You learn conversion intent at the exact moment of friction. The data alone is worth more than a quarter of feature work. SCOPE-01, is this in or out?"
SCOPE-01 (Scope Enforcer)
"Building on METRIC-01's 48-hour trial idea — what if we made it even more aggressive? A 'skip-limit-free weekend' every month. Or a 'bring a friend' trial where both get Premium for a week — you convert two at once and add a social hook. Extending RETAIN-01's music memory layer: what if that annual taste map was generative, not just retrospective? 'Here's your musical DNA — here are 10 artists you've never heard that fit it perfectly.' The memory becomes a discovery engine. And pushing POWER-01's API idea further: a public 'listening graph' — opt-in, anonymized — that third-party apps can build on. Spotify becomes a platform, not just an app."
POWER-01 (Power User)
"I want collaborative playlists with version history — like Git for music. I've built hundreds of playlists over 6 years and I've accidentally deleted things I can't recover. That's Premium-only, zero AI required, pure utility. Also: give me an API. Let me build my own dashboards on my listening data. There's a whole segment of us who'd pay just for that. We exist. We talk about this constantly. Nobody at Spotify has ever asked."
GROWTH-01 (Growth Strategist)
"Wild card round: 1) Spotify for kids as a standalone SKU — parents will pay separately from their own sub. 2) Spotify Live — a ticket-bundled Premium tier: €15/month and one virtual concert included. 3) Spotify Pause — Premium that you can pause for up to 3 months and resume without losing your data or playlists. People cancel because they think there's no other option. Give them the option. You'll recover 30% of churners."
RETAIN-01 (Retention Analyst)
"Building on GROWTH-01's Pause idea — this is the single highest-ROI idea in this session. Every churn study says the #1 reason people cancel is 'I'm not using it enough right now, not that I hate it.' A Pause feature converts a cancel into a snooze. Three months paused, then auto-resume. No re-acquisition cost. This exists for gym memberships. It should exist here."
FREEMIUM-01 (Free-Tier Listener)
"The Pause idea — I'd have used that twice already. I cancelled twice and came back. You could have just paused me. One more: what if free users could 'earn' Premium hours? Listen through an album start-to-finish, get 24 hours Premium. Finish five albums in a month, get a free month. You're rewarding the behavior you want anyway."
CHURNED-01 (Churned Premium User)
"FREEMIUM-01's album-completion idea — I would have stayed for that. Gamification done right: not badges, actual product access. The thing nobody's said yet: Spotify knows when I'm about to churn before I do. Session length drops. Skip rate spikes. Recommendation engagement falls. You have all the signals. Build a 'rescue offer' that fires automatically when the signals hit a threshold — not a discount, a feature unlock. Show me HiFi for a week when I'm drifting. That's when I'm deciding."
Conversion features: 48-hour Premium trial at skip-limit friction point (METRIC-01 — zero engineering, clean intent signal). Social gifting — free users can gift a 3-month trial to one friend per year (GROWTH-01). Milestone unlock — '1,000 hours' page that's Premium-only to share (GROWTH-01). Album-completion earning — finish an album, get 24h Premium (FREEMIUM-01). Event-anchored conversion targeting life moments via listening pattern signals (GROWTH-01). Retention hooks: Pause feature — suspend Premium for up to 3 months, auto-resume (GROWTH-01/RETAIN-01 — highest-ROI idea in session, converts cancel to snooze). Proactive rescue offer firing on churn-signal threshold — feature unlock, not discount (CHURNED-01). 'Taste drift alert' — quarterly musical health check, notification when recommendations plateau (CHURNED-01). One-song check-in at 60 days of no new discovery (UXR-01). Social features: Collaborative listening rooms, real-time shared queue, Premium-only (RETAIN-01). Git-style playlist version history with recovery (POWER-01). Friends family plan — four friends at €15/month total (CHURNED-01). Audio experiences: HiFi unlock as churn-rescue trigger for drifting users (CHURNED-01). Lossless A/B with AI DJ as default home screen for 5% of Premium users (SCOPE-01). 'Spotify Quiet' — no algorithm, no discovery, pure library mode for the leave-me-alone segment (FREEMIUM-01/PMF-01). Liner notes and artist commentary layer (RETAIN-01). Pricing experiments: Spotify for Kids as standalone SKU (GROWTH-01). Spotify Live — ticket-bundled tier (GROWTH-01). Annual music memory review as conversion trigger for long-tenure free users (RETAIN-01). Power user / platform plays: Public listening data API for self-builders (POWER-01 — underserved vocal segment). 'Saved forever' vault vs. free tier playlist expiry after 18 months (RETAIN-01).
Spotify Pause as the single highest-ROI idea — gym memberships have this, Spotify doesn't. The 'rescue offer at churn signal' as a machine learning product that requires no user action. A 'musical genome' reveal at tenure milestones that free users can see but not share.
Prioritize for low-effort/high-signal: METRIC-01's 48h trial experiment (run this week). SCOPE-01's AI DJ home screen A/B (this sprint). RETAIN-01's Pause feature (Q3 scoping). Everything else: backlog for scoring.
UPLOAD YOUR METRICS
Product analytics, cohort tables, churn reports. Pin them as persistent context. Every agent argues with your actual numbers.
CREATE USER PERSONAS AS AGENTS
Your churned users, your power users, your free riders — each one is a character-agent that responds when the council needs their voice.
DELIBERATE PER QUARTER
What to build, what to kill, where to invest. The council debates strategy with evidence from your metrics and direct input from your user-agents.
AGENTS LEARN YOUR PRODUCT
After sessions, GROWTH-01 knows your conversion funnel. RETAIN-01 knows your churn patterns. METRIC-01 remembers which metrics you've already tried optimizing.
HUMAN DECIDES, COUNCIL ADVISES
The synthesis is a recommendation, not a command. You see the evidence, the dissent, the confidence score. You make the call.