# Easy RFP — Social Engine v1.0 (FREE) **Date:** 2026-05-03 **Cost to run:** €0 — runs entirely on your existing Supabase + Cloudflare + Resend stack. **Networks supported:** Facebook (Page), Instagram (Business), LinkedIn (Company Page), TikTok (Business) **Status:** Auto-publish for FB+IG+LinkedIn from day 1. TikTok runs in UPLOAD/reminder mode until TikTok approves `video.publish` scope (4-8 weeks); flips to direct-post automatically after. This bundle ships the complete plumbing to: 1. Schedule posts for all 4 networks from one admin dashboard 2. Auto-publish on the schedule via a Supabase edge function 3. Pull engagement metrics back daily 4. Attribute every signup → paid customer back to the specific post that drove it The killer feature is the attribution chain: post slug → click → cookie → signup → Stripe paid event → MRR. No third-party tool sees your Stripe; no SaaS recurring fee; you own the data. --- ## What's in the bundle ``` social-engine-v1.0/ ├── README.md ← you are here ├── 04-DEPLOY-CHECKLIST.md ← step-by-step deploy guide ├── docs/ │ ├── 00-SETUP-PLAYBOOK.md ← create the 4 business accounts properly │ ├── 01-BRAND-KIT.md ← visual identity, post templates, hashtag pools │ └── 02-UTM-SCHEME.md ← attribution params, anti-patterns, queries ├── migrations/ │ └── 66_social_engine_foundation.sql ← 5 tables + 5 RPCs + 2 views + RLS ├── edge-functions/ │ ├── social-publisher/index.ts ← cron-triggered, posts to FB/IG/LI/TT │ ├── social-metrics-pull/index.ts ← daily metrics pull │ └── social-attribute-signup/index.ts ← Supabase auth hook + Stripe webhook ├── admin-dashboard/ │ ├── index.html ← /app/admin/social/ │ ├── calendar.html ← /app/admin/social/calendar.html │ ├── compose.html ← /app/admin/social/compose.html │ ├── analytics.html ← /app/admin/social/analytics.html │ └── utm-track.js ← drop in public site └── content/ └── 03-CONTENT-CALENDAR.csv ← 35 posts EN/ES/DE ready to import ``` --- ## How it fits into the Easy RFP stack ``` ┌────────────────────────────────────────────┐ │ Cloudflare Pages (static site) │ │ - utm-track.js stamps cookies on landing │ │ - signup form passes UTMs to auth.signUp │ └────────────────┬───────────────────────────┘ │ ▼ ┌────────────────────────────────────────────┐ │ Supabase (existing) │ │ - auth.users (existing) │ │ - Mig 66 NEW: social_* tables │ │ - pg_cron: trigger publisher every 5 min │ │ - pg_cron: trigger metrics-pull daily 6am │ │ - Auth hook → social-attribute-signup │ └────────────────┬───────────────────────────┘ │ service-role ▼ ┌────────────────────────────────────────────┐ │ Edge functions (existing platform) │ │ - social-publisher ← FB/IG/LI/TT APIs │ │ - social-metrics-pull ← FB/IG/LI/TT APIs │ │ - social-attribute-signup ← UTM stamp │ └────────────────┬───────────────────────────┘ │ ▼ ┌────────────────────────────────────────────┐ │ Stripe (existing) → webhook → stamp paid │ └────────────────────────────────────────────┘ ``` --- ## Deploy in 8 steps (~3 hours total) See `04-DEPLOY-CHECKLIST.md` for the granular version. High level: 1. **Apply Mig 66** in Supabase SQL Editor (project `jhihwzfikvndybbyaoul`). Smoke test runs at the end of the migration; you'll see a NOTICE if it passed. 2. **Create the 4 business accounts** following `docs/00-SETUP-PLAYBOOK.md`. ~90 min. 3. **Get API tokens** following the same playbook §6. Save as Supabase secrets: - `META_GRAPH_PAGE_TOKEN`, `META_FB_PAGE_ID`, `META_IG_BUSINESS_ID` - `LINKEDIN_TOKEN`, `LINKEDIN_ORG_URN` - `TIKTOK_ACCESS_TOKEN` - `RESEND_API_KEY` (you already have this) - `SOCIAL_TEST_MODE=true` ← keeps publisher in dry-run until you're ready 4. **Deploy the 3 edge functions** (`supabase functions deploy social-publisher` etc). 5. **Schedule pg_cron** jobs: ```sql SELECT cron.schedule('social-publisher-tick', '*/5 * * * *', $$SELECT net.http_post('https://.functions.supabase.co/social-publisher', '{}'::jsonb, '{"Authorization":"Bearer "}'::jsonb)$$); SELECT cron.schedule('social-metrics-pull-daily', '0 6 * * *', $$SELECT net.http_post('https://.functions.supabase.co/social-metrics-pull', '{}'::jsonb, '{"Authorization":"Bearer "}'::jsonb)$$); ``` 6. **Insert your 4 social_accounts rows** in Supabase (one per network) with `publish_enabled=false` initially. 7. **Drop the admin dashboard** into `/app/admin/social/` of your Cloudflare Pages site (4 HTML files + `utm-track.js` to public assets). 8. **Smoke test**: keep `SOCIAL_TEST_MODE=true`, import the CSV, schedule one post for 2 minutes from now, watch the cron pick it up and "publish" (logs only). Verify no errors. Then flip `SOCIAL_TEST_MODE=false` and `publish_enabled=true` per account, schedule a real post, watch it go live. --- ## What this gives you that SaaS tools (Buffer, Metricool, Publer) don't | Capability | SaaS tools | This engine | |---|---|---| | Auto-post FB/IG/LI | ✅ | ✅ | | TikTok direct-post | ✅ (after their approval) | ✅ (after your TikTok approval) | | Network engagement metrics | ✅ | ✅ | | Attribution to signup | ❌ | ✅ | | **Attribution to PAID customer** | ❌ | ✅ | | **MRR per post slug** | ❌ | ✅ | | Recurring cost | $15-50/mo | **€0** | | Owns the data | ❌ | ✅ | | GDPR audit trail | partial | full (Supabase EU region) | | Fits the existing stack | new vendor | extends what you have | --- ## What this v1.0 does NOT yet do (intentional scope cuts) - **No image generation.** You bring images via Canva or generate elsewhere; the publisher only consumes URLs/storage paths. Auto-generation is a v1.1 feature. - **No A/B testing.** Each post has one variant. Multi-variant ships in v1.1 once you have baseline data. - **No comment/DM monitoring.** Replies on FB/IG/LI/TT don't flow back to the engine. v1.2. - **No Twitter/X support.** Their API is paid + chaotic. Skipped on purpose. - **No native LinkedIn image upload** (v1.0 LI is text + link preview). LinkedIn's 3-step upload flow is in v1.1. - **No threaded posts / TikTok carousels.** Single-post per slug. These are not blockers — the goal of v1.0 is to ship the attribution loop and prove ROI per post. Everything else is incremental. --- ## SOCIAL_TEST_MODE = the kill switch you want Just like `EMAIL_TEST_MODE` in the outbound engine, `SOCIAL_TEST_MODE` is the master safety toggle. - Defaults to `true`. - When `true`, the publisher claims posts but only LOGS what it would have done. No network calls. - Flip to `false` ONLY when you've smoke-tested every account. - Each account also has its own `publish_enabled` boolean — flip those one at a time. --- ## Cost projection at scale | Volume | Supabase cost | Edge fn cost | Storage | Total | |---|---|---|---|---| | 10 posts/day, 4 networks | free tier | free tier | <100MB | **€0** | | 50 posts/day, 4 networks | free tier (500MB DB OK) | free tier (200K invocations) | <500MB | **€0** | | 200 posts/day, 4 networks | might hit Pro €25/mo | possibly Pro €25 | 2GB → free still | €25-50 | Even at 200 posts/day across 4 networks, you'd be at €50/mo total — roughly Buffer's price for a single user, with full attribution and zero vendor lock-in. --- ## Next sprints Once v1.0 is live and you've collected 30 days of data: - **v1.1** — image generation via Canva API or Bannerbear Free, A/B testing, LinkedIn image uploads, more languages. - **v1.2** — comment/DM ingest (DM-to-Slack-via-Resend pipeline), reply classification. - **v1.3** — auto-suggest post times based on per-network audience timezone (you have hotel timezones in `hotels.timezone`; add the same for organizers). - **v2.0** — content engine that drafts posts from product changelog + recent customer wins via Claude API (when ANTHROPIC_API_KEY is wired). --- ## License & reuse Internal Easy RFP tool, but the structure is generic. If you ever want to spin this out as a standalone SaaS, the patterns are clean (compare Outbound Engine sprint 2.x for proof — same shape). Built with the Easy RFP plugin patterns and Supabase + Cloudflare-native conventions. No third-party dependencies beyond the platform APIs themselves.