Building a $9K MRR Design Subscription from $0

Before writing a single line of code, I needed to answer: what does Sako look like? The name 'Sako' came from a brainstorm session — it needed to feel sharp, fast, and B2B credible.

Brand decisions: Logo — green abstract 'S' mark (symbolizes speed, flow, systems). Colors — Neon green (#9AE83E) + purple gradient + dark mode default. Typography — Switzer Variable (clean, modern, B2B SaaS aesthetic). Spacing — 4-point system. Components — buttons, avatars, image placeholders, link styles.

Created a full branding guideline doc in Figma for the team to stay consistent across every surface — product, marketing, decks, social.

Before writing a single line of code, I designed the full landing page in Figma. This prototype validated the visual direction, the narrative structure, and the product story before committing to development.

The prototype established: hero section with the 48-hour SLA promise front and center, 'how it works' flow (submit → draft → review → ship), pricing table with 5 plans, and FAQ that pre-handles objections.

This prototype was used in early sales conversations before the live site was ready — and it closed the first demo calls.

I shipped the full Sako product stack in 3 months using AI-assisted coding (Cursor, v0, Claude). Four apps, one budget, zero compromises on product quality.

Marketing site (trysako.com): 42 pages of MDX content, Figma → Next.js (Tailwind, Framer Motion), OG image generation, sitemap, JSON-LD for SEO.

Client app (app.trysako.com): Dashboard for submitting briefs, tracking tasks, managing brands, Stripe billing integration (checkout, subscriptions, webhooks), review board for asset feedback.

Studio app (studio.trysako.com): Admin kanban for task management, client overview (active clients, MRR, capacity tracking), AI insights (task estimation, complexity scoring).

API (api.trysako.com): NestJS backend (PostgreSQL via Supabase), Stripe webhooks, task workflows, SLA tracking, Composer Gateway (SSO for AI design generator). Total cost: $500.

Tech stack: Next.js 16, NestJS, Clerk (auth), Stripe, Supabase (PostgreSQL), PostHog (analytics), Railway + Vercel (hosting). React 19, TypeScript, Tailwind CSS 4, shadcn/ui, Framer Motion.

I built Composer: an AI-powered landing page generator for Sako clients. The idea — give users a free AI tool, then upsell them to the human pod when they need higher quality.

What it does: Generates landing page copy + layout using LangGraph multi-agent workflows. Routes through OpenRouter (GPT-4, Claude, Gemini). Includes ClaimsOS — a compliance/claims validation system that checks if AI-generated copy makes unsupported claims. 'Send to Sako' button hands AI drafts to the human pod for polishing.

Tech stack: Next.js 16, TypeScript, Tailwind 4, LangGraph 1.x for multi-agent orchestration, Vercel AI SDK 6.x for streaming, Supabase (auth + database), Trigger.dev for background jobs, Bun runtime.

Lesson: PLG doesn't work for productized services. Clients wanted human quality from Day 1, not AI drafts. Composer was discontinued — the human pod was always the real differentiator.

I built a self-hosted n8n instance (n8n.sako.agency on Railway) to automate the entire cold outbound machine — from lead capture to email delivery to Slack-based approvals.

Workflows built: (1) Lead capture + enrichment — Apollo, Clearbit webhooks. (2) ICP scoring — LangGraph agents score leads based on company size, tech stack, and hiring signals. (3) Email sequences — Brevo drip campaigns, warm-open tracking, follow-up triggers. (4) Slack command center — /sako commands for approvals, metrics, workflow actions. (5) Analytics — weekly growth reports. (6) PR outreach — hook generation, tracking.

Integrations: Apollo (sequencing, contact data), HubSpot (lifecycle tracking), Resend/Brevo (email infrastructure), OpenAI/Anthropic/Google (LLM APIs for enrichment). Neon PostgreSQL 17 as the automation database.

Brevo campaign results (17 campaigns total): Peak open rate: 59.49% (EU - Sako, Jan 2026). Consistent open rates: 18–25% across Feb/Mar 2026 campaigns. ~600 recipients per campaign by Feb 2026. Total outbound sequences: January through March 2026.

I designed Sako's pricing strategy and unit economics from scratch — no playbook, just first-principles thinking about what the market would pay and what we could deliver profitably.

Plans: Composer Free ($0/month, AI only), Composer Pro ($30/month, AI + 100 credits), Core ($2,500/month, 1 active task, design + motion, 48h SLA), Ops Base ($4,500/month, 2 active tasks), Enterprise ($8,500+/month, 3+ tasks + strategy).

Unit economics (Core plan): Revenue: $2,500/month. Cost of delivery (freelancer fees, tools, overhead): ~$1,200/month. Gross margin: ~48%. Key insight: productized agencies need 50%+ gross margins to scale profitably.

Entity registration: Registered Sako in the US (Newark, Delaware) via Stripe Atlas to unlock Stripe's full feature set and US payment rails. Also registered a Pvt. Ltd. in Chennai, India, to handle freelancer payments legally. Two entities, one product, $500 total budget.

I personally cold called 700+ B2B SaaS founders and marketing leads in Q1 2026. No SDRs. No warm intros. Just me, a list, and a pitch.

Process: Sourced leads from Apollo, LinkedIn Sales Navigator, and manual research. Built a list of 700 companies (B2B SaaS, $500K–$10M ARR, hiring for design/marketing roles). Called every single one. Pitched Sako's 48-hour SLA. Booked 15+ demos. Closed 2 clients at $4,500/month each = $9K MRR.

Conversion rate: ~0.3% (2 closed / 700 calls). Not great in isolation — but enough to validate the offering. The ones who said yes loved the 48-hour SLA. Speed was the real differentiator.

The pitch deck that closed clients: Designed the client pitch deck in Figma. Every slide was optimized to answer the first objection before it was asked — why not an agency, why not in-house, why now.

The 48-hour SLA was the core promise of Sako. Making it work with a team of 7 remote freelancers required systems, not heroics.

Task workflow: QUEUED → IN_PROGRESS → IN_REVIEW → COMPLETED. SLA timer starts when brief is marked 'Ready' — scope clear, assets provided, one approver assigned. SLA pauses if client delays feedback.

Tools: Studio app (kanban, capacity tracking), Slack (real-time notifications), Review board (in-app asset feedback with annotations), PostHog (product analytics), Resend (email notifications).

Result: Hit the SLA ~85% of the time. Misses were usually client-side delays (late feedback, missing assets). Built SLA tracking directly into the product so clients could see exactly where their task was — and why it might be delayed.

With no budget for paid ads, SEO was the only scalable acquisition channel. I automated 170+ blog posts using AI content generation workflows.

Process: Keyword research (manual + Ahrefs). LangGraph workflows for content generation (GPT-4 + Claude). Automated publishing via Sanity CMS. Internal linking, meta descriptions, OG images. Topics: design ops, productized services, landing page best practices, SaaS marketing.

Result: 170+ posts published in 3 months (Nov 2025 – Jan 2026). Organic traffic: 0 → 500 sessions/month. The blog became the top acquisition channel after cold outbound.

Beyond the product and outbound, I designed Sako's social media presence — visual content that carried the brand into LinkedIn and Twitter feeds.

The SMM content used the same design system as the product — neon green, Switzer Variable, dark mode — so every post felt like an extension of the brand, not a disconnected social team output.

I ran 17 cold outbound email campaigns from Jan–Mar 2026 using Brevo (email infrastructure) + Apollo (contact data). Total recipients across all campaigns: 6,000+.

Campaign strategy: Segmented by geography (US, EU, APAC). A/B tested subject lines on most campaigns. Sent 2–3x per week to different segments.

Best-performing campaigns: (1) EU - Sako (Jan 20): 59.49% open rate, 2.08% CTR, 98 recipients — tight targeting + regional relevance. (2) US - Sako (Jan 20): 36.67% open, 0% CTR, 106 recipients — opens were strong but CTA didn't convert. (3) Feb Outbound 3 (Feb 13): 22.72% open, 1.89% CTR (highest click rate), 480 recipients — 'Book a Fit Check' CTA was the winner. (4) Feb Outbound 9 (Feb 27): 21.64% open, 1.49% CTR, 609 recipients.

Average metrics across all 17 campaigns: Open rate: 18–25% (industry average for cold B2B is 15–20% — we were above average). Click rate: 0.5–1.9% (industry average is 2–3% — slightly below, needed better CTAs). Unsubscribe rate: 0.17–1.28% (low = messaging resonated).

Key insight: Cold email builds awareness, but it's not a closer. The real conversions came from the email → call → demo → close sequence. Email warmed people up; the call closed them.

I tracked all traffic to trysako.com using GA4 (Jan–May 2026). 2,488 active users. 2,486 new users — nearly 100% cold audience, zero brand recognition at the start.

Traffic sources: Brevo email (1,617 users, 65% of all traffic). Direct/typed URL (739 users, 30% — brand recall from cold calls). Organic Google (26 users — SEO just starting after 170+ blog posts). Google CPC (18 users, $100 test budget). LinkedIn referral (8 users).

Top landing pages by active users: (1) Homepage — 1,344 users, 10,800+ views. (2) Fit Check booking page — 779 users, 1,129 views (strong demo intent). (3) Pricing page — 271 users, 337 views (bottom-funnel). (4) ROI Calculator — 146 users, 201 views (highest intent signal, 29.7% bounce rate — lowest on the site).

Bounce rates told the real story: Homepage 51.3% (half engaged). Pricing 39.6% (60% went deeper — strong signal). ROI Calculator 29.7% (highest intent — should have used this as the primary CTA in emails). Product page 23.2% (users who got here were deeply engaged).

Geographic distribution (top cities): Ashburn VA (373, likely AWS infra traffic), Singapore (193), Chennai (104), Melbourne (113), Sydney/Canberra (216 combined). Australia was surprisingly strong — APAC SaaS market responded well.

Key learnings: Email drove 65% of traffic, but only 31% of visitors reached the booking page. The ROI Calculator had the lowest bounce rate — should have promoted it as the primary email CTA instead of 'Book a Demo'. SEO needed 6+ months to compound (only 26 organic users after 3 months of blogging). Direct traffic (739 users) = brand recall from cold calls — people heard the name, then typed it in.