From FP&A Analyst to SaaS Founder: The 3-Month Journey Behind Okiela

Three months ago, I was an FP&A analyst helping Fortune 500 companies model their finances. Today, I am building Okiela -- an AI-powered profit analytics platform that helps e-commerce founders see their real numbers.

This is the honest story of how it happened, what I learned, and what I wish I had known from day one.

The Spark: A Friend's $2M Problem

It started with a phone call. A friend running a Shopify store with $2M in annual revenue asked me a simple question: "What is my actual profit margin?"

After spending 6 hours in Excel pulling apart their Shopify exports, calculating hidden costs, and allocating ad spend per SKU, I had the answer. Their "healthy 55% margin" was actually 28% after all costs.

That is when it clicked: millions of online sellers have this exact problem, and none of them have 6 hours or an FP&A analyst to figure it out.

Month 1: Learning to Build (Zero to Prototype)

I had zero coding experience. No CS degree. No bootcamp. Just a clear problem and determination to solve it.

My tech stack (everything I used to build the first version):

• Cursor AI for code generation and debugging
• v0.dev by Vercel for UI component scaffolding
• Supabase for authentication and database (no backend coding needed)
• Vercel for deployment (push code, it goes live)
• DuckDB WASM for in-browser data analysis

The Key Lesson from Month 1

You do not need to know everything to start. AI tools let you build and learn simultaneously. I was writing React components on day 3, deploying to production by week 2.

The biggest mistake? Trying to build too many features at once. I wasted a full week building a PDF invoice parser before realizing I should focus on one thing: Shopify CSV exports to profit dashboards.

Month 2: The Product Takes Shape

The core insight that shaped everything was the 5D Pipeline concept:

1. 1Upload -- Drop your CSV file
2. 2Clean -- Remove duplicates, fix formatting errors
3. 3Normalize -- Standardize data across platforms (Shopify, Shopee, TikTok Shop)
4. 4Enrich -- Add calculated fields (true profit, margins, ROAS)
5. 5Analyze -- Generate insights and actionable recommendations

The Game-Changer: AI Chat

The feature that transformed the product was adding an AI chat that could answer questions using the actual uploaded data. Not generic advice -- specific, data-backed answers:

• "Which SKU should I reprice?" --> "Raise Jump Rope Speed by $4.75 to hit 30% margin"
• "Why is Travel Kit losing money?" --> "Return rate is 32% and shipping cost exceeds the $2.99 you charge"

This changed Okiela from a "dashboard tool" to a "profit copilot."

Month 3: Launch and First Users

We launched with a simple landing page and a free tier. Key results from the first 30 days:

• Strong early interest from Shopify founders
• Average time to first insight: under 60 seconds
• "Sample Store" demo is our highest-converting feature
• Positive feedback from early users

What Worked

1. 1Start with one platform (we chose Shopify exports first -- 70% of our target users use Shopify)
2. 2Charge from day one (Free tier with 3 analyses/month + Pro at $29/month)
3. 3Talk to users daily (their feedback shaped every feature we built)
4. 4AI tools are a superpower for non-technical founders

What I Would Do Differently

1. 1Start collecting email addresses earlier (we missed 3 weeks of potential newsletter subscribers)
2. 2Add the "Try Sample Store" demo on day one (it is our highest-converting feature)
3. 3Focus on one marketing channel first instead of spreading thin across Twitter, Reddit, and Product Hunt

The Numbers (Transparent)

I believe in building in public, so here are real numbers:

• Total development cost: $0 (all AI tools have free tiers)
• Hosting cost: $20/month (Vercel Pro + Supabase Free)
• Time invested: ~400 hours over 3 months
• Revenue: Early traction, growing steadily

What Is Next

We are building marketplace integrations (Lazada, Shopee) for Q2 2026, and our early adopter program offers a 50% lifetime discount on Pro.

If you are thinking about building a SaaS product, my advice is simple: pick a real problem you deeply understand, use AI tools to build fast, and ship before it is perfect.

The best version of your product is the one your users are actually using.