GTM Plan

Go-to-market spine. Launch sequence, business model, content strategy, support + abuse policy, success metrics. Reads after MANIFESTO.md. The first year, plotted.

Status: locked 2026-04-27 (Phase A pre-launch). v1 GTM plan. Bar: 10/10 across vision + craft + go-to-market. No category falls below A.

1. The plan in one paragraph

Phase A ships single-user (Kevin’s Ensemble dogfoods). Phases B–C deepen the experience while building publicly visible content (showcase pages, the orchestra Studio, the Canvas map) on Kevin’s Ensemble — so by the time we open multi-tenancy, there’s already proof of what it looks like at peak. Phase D opens via a curated 25-person closed beta of hand-picked makers seeded across creative domains. Phase E + open registration follow once the beta is dialed. The first public launch moment is Show HN + Product Hunt + a blog-ready manifesto post, timed when there are 100 active Ensembles with public Studios and the showcase feed has visible quality. Targets: 100 in beta by launch month, 1,000 paid in year one, 10,000 active by year two.

2. Launch sequence

Six waves, sequenced over ~9 months from now. Each wave is gated by a quality bar, not a date.

Wave 0 — Single-user dogfood (Phases A–C, ~6 weeks)

Goal: Phases A, B, C ship on Kevin’s Ensemble. The Studio is dialed. The Canvas map looks great. Showcase pages are ready to share.

Marketing during this wave: - Zero. No tweets, no waitlist page, no announcements. - Dogfood with Kevin and 1–2 trusted reviewers (Owen, possibly an early friend). - Internal notes captured at docs/playtest-notes.md.

Exit criteria: - Kevin has used his own Studio Room daily for 2+ weeks. - 5 built-in presets are all dialed (Apple-quality, no template-feeling). - Sprite library has ~85–100 sprites total (per PRODUCT_SPEC.md §3.3 — a deliberate quality-over-quantity step down from the morning spec’s 80+200 split). - Claude chat in edit mode produces useful output 80%+ of the time. - Showcase pages exist for at least 4 of Kevin’s 8 chats. - Map tab works with a real relates_to graph.

Wave 1 — Stealth waitlist (~2 weeks)

Goal: Build a 200-person waitlist of the right people without ever publicly announcing.

Marketing: - Single landing page at ensemble.tld with: hero (the collective showcase grid in seed form — Kevin’s own canvases plus 4–8 placeholder featured slots filled with the procedurally-generated canvases of pre-onboarded Wave-0 friends, animating live), the manifesto (essay-form, not feature list), a “Get notified” form. No screenshots of features. No feature list. (Wave 1 is pre-Chunk 5; this seed grid is the v1.0 collective-showcase grid in its earliest populated state, not a separate manifesto-only page.) - Page is shared only by Kevin in personal DMs to people whose work he’d want on the platform. Zero broadcast. - Each invite is personal: “I made this with you in mind because of [specific work]. Want early access?” - Goal: 200 hand-picked names, with notes on each (“indie iOS dev, ships with Claude,” “music producer, big Substack”).

Exit criteria: - 200 named waitlist entries. - 10 of those have replied with substantive interest (not just “sign me up”). - Kevin has written follow-up notes to all 200 explaining what’s next.

Wave 2 — Closed beta (Phase D, 25 people, ~6 weeks)

Goal: First 25 multi-tenant users. Each personally invited. They become the showcase + feed seed.

Marketing: - Personal invitations to the 25 highest-fit waitlisters (drawn from the personas in the manifesto §6). - Each invitation includes: their handle reserved, a “first you should…” onboarding doc, a private Slack/Discord for beta feedback. - Phase E (feed) ships during this wave. Beta members are the first feed contributors. Their Studios + showcase pages seed the feed with high quality. - Feedback loop: weekly beta-only office hours over Zoom (Kevin host, ~30 min, 6–8 people each).

Exit criteria: - 25 active users (defined: at least one save in the Studio, at least one public showcase page). - Feed has ~200 events. - Studio reliability >99% (no data loss). - Claude integration works for everyone (no rate-limit bottlenecks). - Net Promoter Score from beta members: 8.5+.

Wave 3 — Soft public launch (~4 weeks)

Goal: First 250 public signups. The platform handles itself without Kevin manually intervening.

Marketing: - The remaining 175 waitlisters get invited. - The blog (at ensemble.tld/journal) goes live with the manifesto + 2–3 case studies of beta users (with their permission). - Twitter / Bluesky presence opens — only sharing beta-user work, not “buy our product.” - Are.na block published at are.na/ensemble. - The waitlist landing page becomes a public homepage.

Exit criteria: - 250 signups. - 100 active (defined: 1+ save in Studio + 1+ public showcase). - Daily active users at ~30% of total (high engagement signal). - Support volume <2 tickets/day (sustainable).

Wave 4 — Show HN + Product Hunt (the launch moment)

Goal: The “we exist” moment. Done deliberately, once everything is ready.

Timing: when 100+ active Ensembles have public Studios visible AND the feed has 1000+ events AND 10+ stand-out showcase pages. Not before. Could be 3–6 months after Wave 3 starts.

Surface: - Show HN post linking to the live homepage + a curated showcase URL. - Product Hunt with a one-image hero (animated GIF of Kevin’s orchestra Studio). - A blog post on the day-of: “Building Ensemble: a year of solo making with Claude” — the founder’s-letter format. 3000-5000 words. Personal, specific, not promotional. - The launch is not the moment we ask for money. We ask for sign-ups, feedback, and curiosity.

Goal: 2,000 signups in 7 days. 500 active. Featured on PH #1–5 product of the day.

Wave 5 — Steady-state growth

Goal: Year 1 close at 1,000 paid users (~10% conversion of 10k signups).

Mechanisms: - Showcase pages naturally rank for project-name SEO. - Studio Rooms are screenshot-able + shareable on Twitter / Bluesky / Are.na. - Each public profile is a referral surface (people see what others built and want to try it). - Featured weekly: one outstanding new Ensemble member highlighted on the homepage + in the feed. - No paid acquisition in year 1.

3. The first 10 hand-picked beta users

Drawn from the personas in the Manifesto §6. These are TARGETS, not commitments. The actual roster is built during Wave 1.

Each gets: - Their handle reserved. - A 1-on-1 onboarding session (30 min over Zoom). - A personalized welcome note in the feed (Kevin’s message visible to them). - Direct Slack/Discord channel. - Free Pro tier for life (the first 25 beta users specifically).

4. Content strategy

What we publish; what we don’t.

4.1 The blog (ensemble.tld/journal)

Cadence: 2 posts/month. 1500–4000 words each. No sponsored content, no roundups, no listicles.

Reference style: The Browser Company’s Notes app, Stripe’s blog circa 2018, Are.na’s editorial.

4.2 Social

• Twitter / Bluesky — share beta-user work. Quote-amplify, never broadcast our own. Personal voice, not corporate.
• Are.na — block called “Ensemble.” Curate showcase URLs from the platform + relevant outside references that show the genre.
• Mastodon — minimal presence. Mirror Bluesky.
• LinkedIn — none. The audience isn’t there for this product.
• Newsletter — Substack at ensemble.substack.com. Two posts/month, mostly mirroring the blog. CTA: try the live thing.

4.3 What we don’t do

• No SEO-bait roundup posts.
• No “X tips for using AI” listicles.
• No comparison posts (“Ensemble vs. Replit”).
• No paid social ads (year 1).
• No press releases.
• No webinars.
• No “AI tool of the day” tweets.

The discipline is: everything we publish is something a thoughtful reader would want to read on its own merit, not because it’s strategically positioned content.

5. Business model

5.1 Free tier

5.1.1 Theme customization (every tier — locked 2026-04-29 evening)

Two-axis theme system available on every paid + free account from v1.0:

• Primary color — pick from 21 themes (the bronze-on-ink default plus 20 alternates spanning warm-to-cool, see ART_DIRECTION.md §2.2).
• Secondary accent — optional override for highlights / CTAs; defaults to primary’s accent.
• Mode — Light / Dark / Auto (Auto follows prefers-color-scheme).

Combined with BYOK, “pick your color, bring your key” is the v1.0 free-tier mechanic — both customisation surfaces are unlocked at signup. See PRODUCT_SPEC.md §3.5 and CHUNK_PREFS_API_CONTRACT.md for the full schema.

5.2 Paid tier — “Studio Pro”

Price: $12/month or $108/year (25% annual discount). Per individual user.

Why $12/month: sits below the psychological “premium subscription” threshold ($15+) but above the “throw-away” threshold ($5). Comparable to Notion ($10), Things 3 (one-time $50, ~$1/month over 4 years), Are.na ($5–$25/yr by tier), Glasp ($10/mo). $108/yr matches the “indie maker $9/month” sweet spot.

Hosted Claude allotment: 500 prompts/month covers 80% of users at average usage. Heavy users hit the cap and either upgrade to add-on packs ($5 / 500 additional) or revert to bring-your-own-key. Cost to us at sonnet 4.7 pricing: ~$2/user/month gross, ~$1/user net after API discounts at scale. Margin: $11/user/month at the $12 tier.

5.3 Team tier (Phase F+)

For organizations using Ensemble: $40/user/month with 5 users included ($200 base), shared workspaces, collaborative Studio editing, admin controls, SSO. Not a Phase A concern; placeholder so we don’t paint ourselves in.

5.4 What we never charge for

• Auth.
• Basic dashboard usage.
• Showcase pages on free tier (within limits).
• Reading the feed.
• Switching plans down (no penalty).
• Procedural empty-state fill. Every signup gets a populated first canvas regardless of tier. The world feels lived-in for everyone; that’s a public good, not a paywall.

5.5 Refund policy

30-day no-questions-asked. Full refund on any paid plan within first 30 days. Pro-rated refund anytime for annual plans (mature operations move).

6. Marketing surfaces

The platform itself is the strongest marketing surface. Each of these is operational from Phase D onward:

6.1 Public profiles

ensemble.tld/<handle> — the user’s public profile page, showing: - Their Ensemble’s Studio Room (live, animated, theme-applied). - Their public showcase pages. - Their recent feed events. - A “Made with Ensemble” badge linking back to the homepage.

Every member is a referral link. Every showcase URL surfaces the platform.

6.2 The collective showcase at homepage hero

The homepage (/index.html) shows the collective showcase grid — a beautiful static grid of canvases from across the community (per PRODUCT_SPEC.md §3.6.6). Pagination, light filtering (newest / featured / random), click any canvas to fly into it. Kevin’s own Studio appears on the grid as one canvas among many — featured by curation, not by being the homepage. When you visit ensemble.tld, you see “the work everyone created together,” not a stock illustration and not a single creator’s Studio.

6.3 Showcase URL embeds

Every public showcase has og:image + open-graph metadata so links shared on Twitter / Bluesky / iMessage render as a rich card with the project’s pie chart, name, phase, and accent color. Free passive marketing.

6.4 Curated weekly highlight

Every Friday at 10am ET: a single Ensemble member featured on the homepage above the fold. Featured member gets a designed badge in the feed + 1k–10k impressions. Earned distribution.

6.5 Genre tags + discovery

Public Ensembles can self-tag with primary genre (music / writing / design / dev / education / research / other). The Discover tab in the feed surfaces by genre. Members find each other within their domain.

7. Safety + trust policy

What we do when bad things happen.

7.1 Content policy

Public showcase pages, Studio Rooms, and feed posts must: - Not violate copyright (especially for sprite-authoring outputs — Claude SVG generations are flagged for derivative-works risk). - Not contain hate speech, harassment, or content targeting individuals. - Not contain CSAM or content sexualizing minors. - Not promote violence or self-harm. - Not impersonate other people or organizations. - Not contain malware or attempt to compromise other users’ browsers.

Reports go to a triage queue. SLA for first response: - Critical (CSAM, malware, doxing): immediate action, content removed within 1 hour, account suspended pending review. - High (harassment, hate speech): action within 24 hours. - Medium (copyright, low-confidence rule violation): action within 72 hours. - Low (style, taste, preference): no action; reported to the user.

7.2 Reporting flow

Every public surface has a Report link in the footer. Reporters describe the issue + cite the offending element. Reports land in a single inbox (initially Kevin; later, a small moderation rotation).

Three actions available to the moderator: 1. No action (with rationale logged for audit). 2. Hide (content unpublished but recoverable; user notified). 3. Remove + suspend (content gone permanently; account suspended pending appeal).

Appeals via email; reviewed within 72 hours.

7.3 Data retention + deletion

• Free / paid users: data retained as long as the account is active.
• Suspended accounts: data retained for 30 days; user can export during this window.
• Deleted accounts: user-initiated deletion hard-deletes within 30 days (for backup recovery), unrecoverable after.
• Reported / removed content: kept in audit log for 1 year, otherwise deleted.

7.4 Privacy-by-default

Restated from the architecture doc: - Every Studio Room is private until explicitly published. - No third-party scripts. - No analytics on user content. - User audio / video / images are user property; never used for training.

7.5 Audit log

Every state-changing action (publish, unpublish, content removal, account suspension, refund) is logged to audit-log/<date>.jsonl with full context. Logs retained for 1 year. Available to user on request (their own actions only).

8. Support

8.1 Channels

• In-app “Need help?” link in Settings. Shows: FAQ → Email → Discord (if subscribed).
• Email: support@ensemble.tld. SLA: ≤7 days free, ≤24 hours paid.
• Discord: invite-only for Pro subscribers. Real-time chat with Kevin + future support team. ~1k users at year 1 = manageable for one person.
• Status page: status.ensemble.tld. Auto-updated by uptime monitor.

8.2 FAQ

Maintained at docs/faq.md (in-product) and ensemble.tld/help (public). Five categories: - Getting started (10–15 questions) - Studio (10 questions) - Privacy + security (5 questions) - Billing + plans (5 questions) - Troubleshooting (10 questions)

8.3 Issue templates

Three intake forms: - Bug report: structured, includes auto-collected version + browser + last-action info (with user consent). - Feature request: free-form, with category tags. - Account / billing: priority routed to Kevin.

8.4 Response posture

Tone matches the product: direct, confident, terse. No corporate-support-speak.

9. Success metrics

What we track. What we ignore.

9.1 The four metrics

These are the only metrics that get reviewed weekly. Everything else is operational color.

9.2 What we don’t optimize for

• DAU/MAU ratio. Ensemble is for daily-but-considered use; not for habit-loop engagement.
• Time-on-site. We’d rather you spend 20 minutes a day in the Studio than 4 hours scrolling the feed.
• Click-through rates on emails / notifications.
• Viral coefficient. We grow on quality, not virality.
• Total signup count. Active users matter; sign-ups don’t.

10. The first year — month-by-month

Month 1 (now → Phase A ship)

• Phase A built (Studio v1).
• Kevin dogfoods.
• Manifesto + ART_DIRECTION + ARCHITECTURE + PRODUCT_SPEC docs final.
• Internal-only.

Month 2

• Phase B (Map tab) built.
• Phase C (Showcase pages) built.
• Kevin’s 8 chats all have showcase pages live.
• Internal-only still.

Month 3

• Quality bar review against the v1 ship list.
• Stealth waitlist landing page live (Wave 1).
• Hand-pick + invite 200 to waitlist.

Months 4–5

• Phase D infrastructure built (multi-tenancy + auth).
• 25 closed beta invites sent (Wave 2).
• Phase E (feed) built during this period.
• Weekly beta office hours.

Months 6–7

• Beta dialed.
• Soft public launch (Wave 3): waitlist opens.
• Blog goes live with founder’s letter + 2 case studies.
• Twitter / Bluesky / Are.na presence opens.

Months 8–10

• 100+ active users sustained.
• 10+ standout showcase pages.
• Wave 4 prep: write the manifesto-class launch post.

Month 11

• Show HN + Product Hunt launch (Wave 4).
• Target: 2k signups in launch week, 500 active by EOM.

Month 12

• Steady-state growth.
• Year-end review: 1,000 paid, 10,000 signups, NPS ≥50.
• Plan year 2 (Phase F + team tier + maybe a desktop app).

11. What this GTM plan is not

• It is not the manifesto — see MANIFESTO.md.
• It is not the product spec — see PRODUCT_SPEC.md.
• It is not the architecture — see ARCHITECTURE.md.
• It is not the art bible — see ART_DIRECTION.md.

It is the people + business spine. How we find users, what they pay (or don’t), how we keep them safe, what we measure. Where this is silent, the manifesto adjudicates.

Locked 2026-04-27. The first 25 beta-user roster gets compiled during Wave 1; the actual launch date is gated by the v1 ship list, not the calendar.