Most marketplace sellers in Turkey think they're profitable. Many of them are wrong.
I'm Serhat Topcu, founder of Netaliz. This is the honest story of building a profit analytics tool for Trendyol sellers: what worked, what surprised me, and what I'm still figuring out.
- Trendyol integration live, Hepsiburada and N11 on the roadmap
- Every sale broken down into 12 separate deduction items
- A demo store anyone can try without signing up
- An AI module that answers customer questions and learns from the seller
1. The problem was hiding in plain sight
Talking to marketplace sellers, I kept hearing the same story: sales were growing, but the money at the end of the month didn't match. They were busy, the orders were coming in, and yet the payout always felt smaller than it should be.
Trendyol's seller panel shows the commission. But the real cost of a sale includes shipping, service fees, withholding tax, VAT, returns, damaged goods and ad spend. Each has its own formula.
Sellers were trying to track this in Excel. Most of them gave up halfway.
Takeaway: The best problems are the ones people already try to solve with spreadsheets.
2. "Just calculate the profit" is never simple
I thought profit was one formula. It's twelve.
Commission isn't even a fixed number. It changes with the product's price, across multiple tiers. Raise the price by a small amount and you might drop into a different tier and earn less.
So a simple "multiply by the commission rate" approach was wrong. For price suggestions, we ended up solving it numerically: give a target margin, and the system searches for the price that actually hits it, tiers included.
Takeaway: If your domain has edge cases, they aren't edge cases. They're the product.
3. Marketplace APIs have personalities
Integrating with a marketplace API teaches you humility. Timestamps don't always line up the way you expect. Commission data changes. Returns arrive long after the original order.
The timestamp issue was a good example. An order could appear on one day in our reports and a different day in the seller's panel, depending on how the time was interpreted. To a seller, that doesn't look like a timezone detail. It looks like the numbers are wrong. We ended up adding in-app guides that explain these marketplace quirks right where they show up.
Our first sync ran on a schedule every 15 minutes. It worked, but sellers watching a live campaign wanted to see sales now. So we added webhooks for near real-time updates, with the scheduled sync kept as a safety net.
Takeaway: Never trust a single data path. Build a backup sync from day one.
4. Trust is a feature
Sellers were giving us access to their store. That's a big ask.
One of the first questions sellers asked was simple: "Can this break anything in my store?"
So we made a deliberate decision: Netaliz is read-only. It never changes products, prices or stock. API keys are encrypted with AES-256 envelope encryption, and data is hosted in Turkey.
We put this right on the landing page, not hidden in a policy document, because that question deserves a clear answer before anyone signs up.
Takeaway: In B2B, what you won't do can sell better than what you will.
5. Let people see it before signing up
People hesitated to connect their real store just to "try" a tool.
The fix was a demo store with sample data and the real interface, no signup required. Visitors can click around, see a red "loss" order, and understand the value in seconds.
Takeaway: Remove every step between curiosity and the "aha" moment.
6. AI was the feature sellers didn't ask for, but use daily
Sellers receive a constant stream of customer questions. Answering them well takes time.
We built Netaliz AI to draft replies using the product's details, the seller's brand voice and their own rules. Every answer the seller approves is stored and used as an example for similar future questions, so it gets better over time.
Nothing is sent automatically by default. The seller reviews and clicks send.
Takeaway: Human-in-the-loop isn't a limitation. Every approval is free training data.
7. What I'm still figuring out
Distribution. Building the product turned out to be the easier part.
Marketplace sellers are busy people. They don't browse software directories or read tech blogs. They live in seller groups, forums and WhatsApp chats, and they trust recommendations from other sellers far more than any ad.
Reaching them in a way that feels helpful rather than salesy is the problem I'm working on right now. Expanding to Hepsiburada and N11 is the other big challenge, since every marketplace has its own fee structure and its own API quirks.
What I'd do differently
- Launch the demo store much earlier. It answers questions no landing page can.
- Start marketing before the product feels "finished". It never will.
- Talk to sellers before writing a single formula.
If you want to see what it looks like, here's the demo store with sample data, no signup needed: netaliz.com/demo
I'd love to hear from anyone building for a local market or integrating with marketplace APIs. What surprised you the most?
Top comments (0)