E-commerce Market Entry Into China: Lessons From a Zero Start

I led the launch of a Western brand into the Chinese market from zero — no presence, no partnerships, no existing demand. It remains the project that most thoroughly rearranged my assumptions about what "international expansion" means, because almost nothing transferred. Not the channel playbook, not the content, not the funnel math, not even the definition of a website. What follows is what I would tell an operator considering the same move: the structural realities first, then the sequence that worked, then the mistakes I either made or watched others make at close range.

Structural reality one: your stack doesn't exist there

The first planning meeting for a China entry should begin by crossing out the tools on your architecture diagram. Google — search, analytics, fonts, maps — is not part of the consumer's world. Neither are Instagram, Facebook or the email-centric CRM motion Western DTC is built on; email open rates in China are a rounding error compared to what WeChat delivers. The ecosystem runs on its own axis: WeChat as identity, messaging and storefront at once; Tmall and JD as the marketplaces where legitimacy is established; Douyin (the domestic TikTok) and Xiaohongshu as the discovery layer where purchase decisions are actually formed. Your existing funnel does not get translated — it gets re-derived from local behavior. Budgeting a China entry as "the US playbook plus translation" is the original sin of the projects that fail quietly two years in.

Structural reality two: cross-border first, entity later

The regulatory question — import licensing, product registration, a local legal entity — looks like the first gate, and for some categories it genuinely is. But China maintains a cross-border e-commerce channel precisely for testing foreign brands: marketplace stores that ship from bonded warehouses or abroad under a distinct customs regime, without a domestic entity. Starting cross-border converts an existential legal project into a bounded operational one, and — more valuable — it generates real demand data before the expensive commitments. The full entity, domestic inventory and general-trade import made sense for us only as a response to proven traction, not as a prerequisite for seeking it.

Structural reality three: trust is imported differently

A Western brand arrives with zero search presence and zero word of mouth, and Chinese consumers are — rationally — deeply skeptical of unknown foreign products, having watched waves of grey-market and counterfeit goods. Trust gets built through channels that look inefficient to a Western marketer: presence on the marketplaces whose buyer protections consumers rely on; reviews and seeded content on Xiaohongshu, where users research brands the way Westerners use search; key-opinion-leader partnerships that function less like influencer ads and more like retail distribution with a face. We learned to treat KOL spend as a channel cost with negotiated deliverables and tracked sell-through, not as PR — the discipline of reconciling every channel's claims against your own numbers applies double in a market where you cannot read the platforms' native dashboards without help.

The sequence that worked

1. Demand evidence before any build. Category research on Tmall and Xiaohongshu: who sells the adjacent product, at what price architecture, with what review language. The review language matters most — it tells you the actual purchase criteria, which in our category differed sharply from the Western ones.
2. Localization as product work, not translation. Naming that survives Mandarin phonetics, claims rewritten for local regulation and local skepticism, packaging and gifting conventions, price points aligned to the festival calendar. We treated the Chinese offer as a new product that shared a supply chain with the original.
3. One channel to credibility. A single marketplace flagship plus one discovery platform, run properly with local-language customer service meeting local response-time norms — instead of a thin presence everywhere. Depth reads as legitimacy; breadth reads as a grey importer.
4. Operations sized for the calendar. Chinese e-commerce concentrates into festival peaks — Singles' Day above all — that make Black Friday look gentle. Inventory positioning, logistics capacity and support staffing get planned backward from those dates. A stockout in peak week does not defer revenue; it donates your search ranking to a competitor.
5. A local operating partner, scoped precisely. A TP (Tmall partner) agency or local team is close to unavoidable for store operations, but the scope needs the same treatment as any vendor: owned accounts, owned data, defined SLAs, and reporting you can audit. Brands that hand the whole motion to a partner discover later that they own a trademark and nothing else.

Mistakes worth skipping

• Assuming the funnel math transfers. Western channel economics did not survive contact; every acquisition-cost assumption had to be rebuilt from local data. Enter with hypotheses, not forecasts.
• Translating content instead of re-deriving it. Our best-performing Western creative was irrelevant; content that worked was made in-market, in the platforms' native formats, by people who lived in them.
• Underestimating service norms. Response-time expectations on Chinese platforms are minutes, not hours, and reviews punish misses immediately. Support is a ranking factor, effectively.
• Treating it as a side project. The entry consumed senior operating attention for far longer than planned. A market this different does not run on a Slack channel checked twice a week — this is one context where I tell founders plainly that advice-shaped help is not enough; someone senior has to own it.

Whether you should

The honest screening question is not "is China big" — it is whether you have a product with a demonstrable local purchase driver (imported provenance that matters in your category, a genuine gap on the local shelf), the working capital to fund a peak-season inventory calendar, and the leadership bandwidth to run what is effectively a second company for at least a year. Two of three is a no. What the project gave me, beyond the market itself, was a permanent upgrade in operating humility: every assumption I carried was tested, most broke, and the ones that survived — instrument everything, own your data, build operations one size ahead of demand — turned out to be the ones that were never about geography in the first place.