Cross-Border E-Commerce from Turkey in 2026: The Technical Infrastructure
Turkey's e-export climbed from 2.2 billion dollars in 2022 to 6.4 billion in 2024, with an 8 billion target for 2025. Most brands that try it fail on infrastructure rather than demand: single-currency pricing, machine-translated pages, a payment stack that cannot take foreign cards, and no answer for VAT or returns. Here is the build that works.
— ByHalil Berkay SahinThe demand side of Turkish e-export is not in dispute. Official figures track the climb from 2.2 billion USD in 2022 (0.90% of general exports) to 5 billion in 2023 (2.12%) and 6.4 billion in 2024 (2.6%), with the Ministry of Commerce setting an 8 billion target for 2025. More than 11,500 businesses cleared 10,000 USD in e-exports. Turkish products in textiles, home goods, jewellery, cosmetics and food have genuine international pull, and the cost base is favourable. What kills most attempts is not demand. It is that a store built for the domestic market is structurally unable to serve a foreign one, and the gaps only show up after the ad spend starts.
Gap one: currency. Displaying prices in lira to a buyer in Germany is a conversion killer even when the arithmetic is fine, because the buyer cannot evaluate the price without doing work. Real multi-currency means price display in the buyer's currency with locale-correct formatting (1.299,00 EUR in Germany, €1,299.00 in Ireland), pricing rules per market rather than a single live FX conversion that produces prices like 43.71 EUR, and honesty about landed cost. Surprise customs duty at the door is the single most reliable way to generate a chargeback and a one-star review simultaneously. Decide explicitly whether you ship duty-paid or duty-unpaid and say so on the product page, not in a footer link.
Gap two: payment. Domestic Turkish payment infrastructure is excellent for domestic cards and largely irrelevant for a buyer in Amsterdam. Cross-border you need an international acquirer path: Stripe for physical goods in most markets, or a merchant of record such as Lemon Squeezy if you sell digital products and want the VAT liability handled for you. That distinction matters more than people expect. A payment gateway leaves EU VAT registration and remittance your problem across every market you sell into. A merchant of record becomes the seller of record and handles it, which for a small digital-goods business is often worth the higher percentage. Run both rails: domestic infrastructure for Turkish buyers, international for everyone else, with routing based on the buyer rather than a single compromise setup that underperforms in both directions.
Gap three: content. Machine-translated product pages are detectable by buyers and increasingly by search engines, and in categories where trust drives purchase, detectably foreign copy suppresses conversion before it ever reaches the checkout. The build requirement is per-market content, natively written, with correct hreflang annotations including x-default so search engines serve the right version rather than picking one. This is also where AI search visibility is won or lost: assistants quoting your product need clean, server-rendered, structured content in the language of the query. Machine-translated pages get read, evaluated as low quality, and skipped.
Gap four: returns, which is where cross-border unit economics actually live. A return from Germany to Istanbul can cost more than the product margin, so the policy has to be designed rather than copied from your domestic store. The options are a local return address in the target region with periodic consolidation, a returnless refund threshold below which you simply refund and do not ship back, or a partner service handling regional returns. Whichever you choose, model it before launch. Fashion return rates that are survivable domestically become the entire margin when each return carries international freight.
Gap five: tax and compliance per market. EU distance selling has VAT obligations, the UK has its own post-Brexit rules, and the US has state-level sales tax nexus rules that surprise almost everyone. This is precisely the calculation behind choosing a merchant of record for digital goods. For physical goods, you need HS codes on your catalogue, correct customs documentation generated automatically rather than typed per order, and an accountant who has actually done cross-border. As with anything tax-related in this article: this is a summary to help you ask better questions, not tax advice. Get a specialist.
The marketplace question, honestly. Etsy, Amazon and eBay solve gaps one, two, four and five for you, in exchange for roughly 15-20% all-in on most category and fee combinations, plus the structural cost that the customer belongs to the platform. For a brand entering a new market with no audience, that trade is usually correct at the start. The mistake is treating it as permanent. Once you have repeat international buyers, you are paying acquisition rates on demand you generated yourself, which is the same trap domestic sellers hit with Trendyol and Hepsiburada, just in a different currency.
What the technical build looks like when done properly. A server-rendered storefront with per-locale routes and complete hreflang including x-default. Market-based pricing rules rather than live FX conversion, with locale-correct number and currency formatting. Dual payment rails routed by buyer geography. Structured product data (Schema.org Product with offers, availability and per-market currency) so both search engines and AI assistants can read the catalogue without executing JavaScript. Shipping and customs documentation generated from the order rather than assembled by hand. And per-market content managed as first-class content, not as a translation layer bolted onto a Turkish original.
How to sequence it without burning capital. Do not launch six markets at once. Pick one target market where your product has a genuine advantage, build the full stack for that single market, and prove the unit economics including returns across at least one full season. Then add markets, because the second market costs a fraction of the first: the infrastructure exists and only content, pricing rules and tax handling are incremental. Brands that launch broad tend to discover their return economics in three markets simultaneously, which is an expensive way to learn one lesson.
The strategic read on 2026. Turkey's e-export volume is growing fast in official statistics but from a base that is still small relative to the domestic market, and government targets exceed realised figures, which tells you the field is not yet crowded with well-built international operations. Most Turkish exporters selling online are doing it through marketplaces with domestic-grade infrastructure underneath. A brand that builds a genuine multi-market storefront, with honest landed cost, native content and designed return economics, is competing against a field that mostly has not made that investment. That window does not stay open indefinitely, but in 2026 it is open.
- 01Official e-export series: 2.2 billion USD in 2022, 5 billion in 2023, 6.4 billion in 2024. The 8 billion figure for 2025 is a Ministry target, not a realised result. Over 11,500 businesses cleared 10,000 USD in e-exports.
- 02Failures are infrastructural, not demand-driven. Domestic-grade stores structurally cannot serve foreign markets, and the gaps only appear after ad spend starts.
- 03Multi-currency means market-based pricing rules with locale-correct formatting and honest landed cost, not a live FX conversion producing prices like 43.71 EUR. Surprise customs duty produces chargebacks and one-star reviews together.
- 04Run dual payment rails: domestic infrastructure for Turkish buyers, Stripe or a merchant of record (Lemon Squeezy for digital goods, which assumes the VAT liability) for international, routed by buyer geography.
- 05Return economics decide cross-border viability. A German return to Istanbul can exceed product margin; design a regional return address, a returnless-refund threshold, or a partner service before launch, not after.
- 06Per-market content must be natively written with complete hreflang including x-default. Machine translation suppresses conversion and gets skipped by AI assistants quoting products.
- 07Launch one market fully and prove unit economics across a season before adding more. The second market costs a fraction of the first; launching broad means discovering return economics in three markets at once.
How big is Turkey's e-export market?
Official figures track 2.2 billion USD in 2022 (0.90% of general exports), 5 billion in 2023 (2.12%) and 6.4 billion in 2024 (2.6%), with the Ministry of Commerce targeting 8 billion for 2025. More than 11,500 businesses recorded over 10,000 USD in e-exports. Note that the 8 billion figure is a target rather than a realised outcome; treat targets as intent, not data.
Should I sell internationally through Etsy and Amazon or my own store?
Both, in sequence. Marketplaces solve currency, payment, returns infrastructure and much of the tax burden for roughly 15-20% all-in on typical category and fee combinations, which is usually the right trade when entering a market with no audience. The mistake is treating it as permanent: once you have repeat international buyers, you are paying acquisition rates on demand you generated yourself. Use the marketplace to acquire and your own store to retain.
What payment setup does cross-border selling from Turkey need?
Dual rails routed by buyer geography. Domestic Turkish payment infrastructure for Turkish buyers, and an international path for everyone else: Stripe for physical goods in most markets, or a merchant of record such as Lemon Squeezy for digital products. The merchant of record distinction matters because it becomes the seller of record and assumes VAT liability across markets, whereas a plain gateway leaves registration and remittance as your obligation everywhere you sell.
How do I handle returns for international orders?
Design the policy rather than copying your domestic one, because a return from Germany to Istanbul can cost more than the product margin. The three workable structures are a local return address in the target region with periodic consolidation, a returnless-refund threshold below which you refund without shipping the item back, and a partner service handling regional returns. Model return costs before launch; fashion return rates that are survivable domestically can consume the entire margin when each return carries international freight.
Is machine translation good enough for international product pages?
No, on two fronts. Buyers in trust-driven categories detect translated copy and it suppresses conversion before checkout. And AI assistants quoting products favour clean, server-rendered, natively written content in the language of the query, so machine-translated pages get evaluated as low quality and skipped. The requirement is per-market content written natively, with correct hreflang annotations including x-default so search engines serve the right version instead of picking one themselves.
Let's apply this to your project.
A free 15-minute call. A technical assessment, not a sales pitch.