Independent site or marketplace: the real cost and payoff of DTC for exporters

Almost every founder we meet who is going global asks the wrong version of this question. They ask "Amazon or my own site?" as if it is a religion. The honest answer is that a marketplace and an independent site do two completely different jobs, and the smart move is knowing which job you are paying for at each stage. A marketplace rents you demand. An owned site builds you an asset. Both cost money. The difference is what is left in your hands after the money is spent. So let us put real numbers on the table, walk through the math the way an operator actually has to, and give you a framework you can apply to your own product this week.

This is a long guide on purpose. The "DTC versus platforms" debate is full of slogans, and slogans are exactly what get exporters into trouble. "Own your customer" is true and useless if your unit economics cannot carry the acquisition cost. "Amazon takes too much" is true and useless if your product has no repeat purchase to monetize an owned relationship. The goal here is to replace the slogans with a model: how to read the fee structure of each channel, how to test whether your product can earn its own demand, how to sequence the two channels over time, and how the rules changed in 2025 in a way that quietly tilts the table. Read it once with a notepad, and you should be able to tell which stage your brand is in and what to do next.

The framing mistake almost every exporter makes

The first thing to fix is the question itself. "Should I sell on Amazon or build my own store?" treats the two as substitutes, as if picking one means rejecting the other. They are not substitutes. They are different tools for different jobs, and a healthy brand usually uses both, just in a deliberate order and for deliberate reasons.

A marketplace is a demand utility. You plug in, and intent-rich shoppers who were already going to buy a product like yours can find and buy yours. You pay for that the way you pay for electricity: per unit, every unit, forever, with no equity in the grid. An independent site is the opposite. The infrastructure is cheap, but it arrives empty. Nobody walks in unless you bring them, and bringing them costs real money. What you get in return is something the marketplace will never sell you at any price: the customer relationship, the first-party data, and the brand surface where margin and loyalty actually accumulate.

Once you see them as two jobs rather than two religions, the real questions get sharper. Which job do I need most right now? Can my product economics carry the cost of the job I want? And what is the right sequence to move from one to the other without going broke in the gap? The rest of this guide answers those three questions in order.

What a platform actually charges you

A marketplace feels cheap because there is no upfront build. The cost shows up later, as a slice of every order. On Amazon, the referral fee is roughly 15 percent of the sale price in most categories, and that is before fulfillment. Once you add Fulfillment by Amazon, total fees on a typical small, light item now land around 31 to 33 percent of the price, and they have been climbing year over year. Then comes the part nobody puts in the deck: to actually get seen, you bid on Sponsored Products against your own category, so a chunk of your "organic" marketplace turns into paid media too.

Add it up and a category leader can be handing 35 to 45 percent of revenue to the platform once ads are included. You are renting traffic. The moment you stop bidding, the traffic stops.

The fee stack, layer by layer

It helps to stop thinking about "Amazon's fee" as a single number and start thinking of it as a stack. Each layer is justifiable on its own, and that is exactly why the total sneaks up on you. The referral fee is the rent for the storefront. The fulfillment fee is the warehouse, the picking, the packing, and the shipping. The storage fee is the rent on the shelf your inventory sits on, and it spikes in the fourth quarter and again if your goods age past six months. Returns processing is its own line. And advertising is the layer that has grown fastest, because as more sellers crowd a category, the only way to stay above the fold is to outbid the seller next to you. None of these layers is unfair. Stacked together, though, they routinely consume a third to nearly half of the sale price, and the seller often does not feel it until the quarterly statement lands.

The fee you cannot see: the data you never collect

There is a cost on the marketplace that never appears on any invoice. When a customer buys from you on Amazon, you do not learn who they are. You get an order, a shipment, and a payout. You do not get an email you can market to, a profile you can segment, or a clean way to bring that person back next month at near-zero cost. The platform keeps the relationship. For a one-time purchase that is fine. For anything with a second order in it, that invisible cost is often larger than every visible fee combined, because it caps your ability to ever earn a customer's lifetime value rather than just their first transaction.

What an owned site actually charges you

An independent store flips the math. The platform cost is small and fixed: Shopify runs from $39 a month on the Basic plan, with payment processing around 2.4 to 2.9 percent plus 30 cents per transaction and apps that typically add $50 to $150 a month. That is your "rent." The expensive part is getting a stranger to arrive at all.

This is where most 出海 brands underestimate the bill. Customer acquisition cost on paid social has risen hard: Shopify's own merchant data shows CAC climbing to around $318 on average across millions of merchants, Meta CPMs hit record highs, and Google Shopping CPCs jumped more than 30 percent in a single year. If you are spending $40 to acquire a customer who places one $55 order, your "owned" channel is quietly less profitable than the marketplace you were trying to escape.

The cheap part and the expensive part

Founders fixate on the cheap part because it is the part with a visible price tag. A theme, a domain, a payment app, a few plugins, and you have a store for less than the cost of a single Amazon storage spike. That is genuinely cheap, and it is genuinely a trap, because it makes the whole channel feel affordable when the real bill is hiding in the line you have not budgeted for: traffic. On the marketplace, traffic is included in the fee. On your own site, traffic is a separate purchase you make over and over, and the price of that purchase has been rising across every paid channel for years.

Why acquisition cost keeps climbing

It is worth understanding why, because it tells you where the durable edge is. Paid acquisition is an auction, and the auction has gotten more crowded. More brands are bidding for the same attention on the same handful of platforms, privacy changes have made targeting blunter, and the platforms themselves are incentivized to let prices rise. The result is structural, not cyclical. If your entire growth plan is "buy ads on Meta and Google," you are buying into the one input whose price is most reliably going up. The brands that win on owned channels are the ones that build sources of traffic that are not an auction at all: organic search, AI answer engines, earned media, and a customer base that comes back without being re-bought.

The two cost structures side by side

Numbers make this concrete in a way that prose cannot. The table below models the same hypothetical brand, a small consumer product with a $50 average order value selling $50,000 of goods in a month, on each channel. Treat every figure as typical and illustrative rather than a promise. Your category, your weight, your margin, and your ad efficiency will move these meaningfully. The point is the shape of the two columns, not the decimals.

Read the bottom two rows again, because that is the whole argument. On the marketplace, the day you stop paying, the demand evaporates and you have nothing to show for the spend except past revenue. On the owned site, even when you pause advertising, the email list, the brand recognition, and the search rankings you built keep working. One channel is an expense. The other is, when run well, an investment that compounds.

The number that decides everything: LTV to CAC

Here is the test we run before recommending DTC to any exporter. Take the lifetime value of a customer and divide it by what it costs to acquire one. A widely used benchmark is that a 3 to 1 ratio is the minimum for healthy ecommerce. A $200 acquisition cost is a disaster against a $220 lifetime value, and it is excellent against a $1,000 one.

This single ratio explains why some products belong on a marketplace and some belong on an owned site. If your product is bought once, rarely, with no natural reorder (a suitcase, a one-time gadget), you will struggle to earn back a high acquisition cost on your own site, so a marketplace that hands you cheap, intent-rich traffic is often the rational choice. If your product gets reordered (skincare, supplements, pet food, coffee, anything consumable) or carries high margin and a community, every owned customer keeps paying you. That is when building your own channel turns from a cost into a compounding asset.

How to actually calculate your LTV

Most exporters guess at lifetime value, and the guess is almost always too optimistic. A workable version is simple. Take the average order value, multiply by your gross margin to get the profit per order, then multiply by the number of orders an average customer places before they stop buying. A customer who spends $50 at a 50 percent margin and buys four times over two years is worth $100 in gross profit, not $200 in revenue. Use gross profit, never revenue, because you can only spend profit on acquisition. If you do not yet have repeat-purchase data, model it conservatively and revisit it once you have ninety days of real cohorts, because the early number is a hypothesis, not a fact.

The payback period nobody talks about

The ratio tells you whether a customer is worth more than they cost. The payback period tells you whether you will survive long enough to find out. A 3 to 1 lifetime ratio is healthy, but if it takes eighteen months to recover the acquisition cost, you have to fund eighteen months of negative cash flow on every customer you acquire. For a venture-backed brand that may be fine. For a self-funded exporter, it can be fatal. A common rule of thumb is to recover acquisition cost within the first ninety to one hundred and twenty days. If your payback is longer than that, slow your acquisition spend until your retention or margin improves, no matter how attractive the lifetime ratio looks on paper.

Two products, two verdicts

Consider two real-feeling examples. The first is a $39 travel adapter. People buy one, it lasts for years, and they have no reason to come back. Even a modest $25 acquisition cost is hard to justify against a single low-margin sale, and there is no second order to rescue the math. This product belongs on the marketplace, where the platform hands you a shopper who already typed "travel adapter" into the search bar. The second is a $45 specialty coffee subscription with a 60 percent margin and customers who reorder monthly for over a year. A $40 acquisition cost looks scary on the first order and looks brilliant by the third. This product belongs on an owned site, where you can capture the relationship and monetize it again and again. Same starting price, opposite verdict, and the ratio is what tells you which is which.

What Anker teaches Chinese exporters

The most useful case study for any 出海 founder is Anker, because it grew up Amazon-native and then deliberately walked away from dependence. In its early years roughly 80 percent of revenue came from Amazon. The company then built out anker.com, soundcore.com and its other brand sites, pushed into 40,000-plus retail doors, and used owned channels for exclusive launches and loyalty so it could capture first-party data and protect margin. The platform stayed a major channel, it just stopped being the only one.

The lesson is not "abandon Amazon." It is sequence. Use the marketplace to prove demand and generate cash flow with low risk, then use that cash to build an owned channel that the platform can never take away from you.

Why concentration on one channel is a hidden risk

The strategic reason to diversify is not just margin. It is fragility. A brand that earns most of its revenue from a single marketplace lives at the mercy of that marketplace's decisions. An algorithm change, a category fee increase, a listing suspension, or a sudden flood of cheaper competitors can erase a quarter of revenue overnight, and you have no relationship with the customer to fall back on. Anker's move into owned channels and offline retail was, in part, an insurance policy. Every dollar of revenue that comes through a channel you control is a dollar that cannot be switched off by someone else. For exporters who have watched accounts get suspended for reasons that were never fully explained, that insurance is worth real money. We have written separately about the worst version of that scenario in Amazon account ban recovery, and about the broader trap of over-relying on marketplaces in China brand vs marketplaces.

The 2025 tariff shock changed the math

There is a new reason this matters more than it did even two years ago. The United States ended the $800 de minimis duty-free exemption, with shipments from China and Hong Kong losing eligibility on May 2, 2025 and the exemption closing globally on August 29. The ultra-cheap, ship-direct-from-the-factory model that powered Temu and Shein got expensive overnight, with duties and flat per-parcel fees landing on goods that used to enter free.

For brands, that does two things. It erases the price advantage of pure cross-border arbitrage, and it rewards companies that hold U.S. inventory, build a real brand, and earn repeat customers rather than chasing the lowest landed cost. In other words, it tilts the table toward owned demand and durable brand equity, which is exactly what a well-run DTC channel is built to create.

What the end of de minimis means for your landed cost

Practically, the change punishes the smallest, cheapest, most frequent parcels the hardest, because a flat per-parcel fee is a much bigger percentage of a $12 order than a $120 one. The model that sent a single low-value package directly from a Chinese warehouse to a U.S. doorstep, duty-free, is the model that just got taxed. The alternative, holding inventory in U.S. warehouses and shipping domestically, now looks more competitive than it did, and it also happens to deliver faster shipping, easier returns, and a better customer experience. If you are weighing fulfillment models, we go deeper on the tradeoffs in overseas warehouse and logistics traps, and on the customs and duty side in China export compliance traps.

Why this rewards brand, not arbitrage

When the price advantage of cheap cross-border shipping disappears, the brands left standing are the ones that gave customers a reason to buy beyond price. A customer who chose you because you were the cheapest option will leave the moment someone is cheaper, and after the tariff changes, the cheapest option is no longer the China-direct parcel. A customer who chose you because they trust your brand, recognize your name, and had a good experience last time is a customer you can keep. That trust is built on owned surfaces: your site, your content, your reputation in search and in the press. The tariff shock did not create the case for brand-building. It just raised the cost of not doing it.

A practical sequence we recommend

For most Chinese brands entering the U.S., the answer is not either-or. It is staged. The mistake is treating one channel as a substitute for the other. They are a relay, not a rivalry. Here is the sequence we walk clients through, in order.

1. Validate on the marketplace. Launch on Amazon to test product-market fit and pricing with minimal infrastructure. Treat the fees as the cost of fast, honest market feedback. You are buying certainty here, and 15 percent is a fair price for learning what sells before you commit to building anything.
2. Build the owned foundation in parallel. Stand up a credible, fast, conversion-focused site so customers who discover you anywhere have a branded home to land on. A weak site quietly caps everything upstream of it. This is the unglamorous step founders skip, and skipping it is why so much ad spend leaks away.
3. Earn free traffic instead of only renting it. Invest in search and AI visibility so customers find you in Google and in answers from ChatGPT, Gemini and Perplexity, lowering blended acquisition cost over time. This is the layer that turns DTC from an expensive habit into a compounding asset.
4. Migrate repeat customers to direct. Use inserts, loyalty and email to move your best customers off the platform and onto a relationship you control. The marketplace found them; your job is to keep them.

Do this and you get the best of both: the marketplace funds the early months, and the owned channel becomes the place where margin and brand equity actually accumulate. Each step funds the next, so you are never betting the company on a channel you have not proven yet.

The bridge step everyone forgets

There is a step between two and four that most brands miss entirely, and it is the cheapest growth lever you will ever pull. When a customer buys from you on the marketplace, you cannot get their email, but you can put a card in the box. A small insert that offers a genuine reason to visit your site, a registration for a longer warranty, a how-to guide, a discount on the next order, quietly moves your best customers from a relationship the platform owns to one you own. It costs pennies per order and, done consistently, it is how a marketplace-native brand slowly builds a direct customer base without paying acquisition cost twice. Treat every marketplace order as a paid introduction, and your job is simply to turn the introduction into a relationship.

A decision framework you can apply this week

If you want a faster answer than reading the whole guide twice, run your product through these five questions. They are the same ones we use in the first hour of a client conversation, and they will get you most of the way to a defensible decision.

1. Does the product get reordered? If yes, an owned channel can monetize the relationship and DTC starts to make sense. If it is a one-and-done purchase, the marketplace is usually the rational home.
2. What is your gross margin after landed cost? Thin margins cannot fund rising acquisition costs. You generally want comfortably above 50 percent gross margin before an owned channel can carry paid acquisition on its own.
3. Can your unit economics hit 3 to 1 LTV to CAC? Estimate honestly using gross profit, not revenue. If the math does not clear the bar, fix retention or margin before you scale spend.
4. Is there demand that already exists to capture? If shoppers are actively searching for your category, the marketplace and search can hand you intent cheaply. If you are creating a new category, you will have to fund demand generation, which is slower and more expensive.
5. How much channel risk can you tolerate? If a single suspension would threaten the business, the case for an owned channel as insurance gets stronger regardless of the pure economics.

Score those five and the pattern usually becomes obvious. Reorderable, high-margin, defensible products lean toward owned channels. One-time, thin-margin, commodity products lean toward the marketplace. Most brands sit somewhere in between, which is exactly why the staged relay above beats a binary choice.

Common mistakes and pitfalls

Over many conversations with exporters, the same expensive errors recur. None of them are about effort. They are about reading the cost structure wrong, and they are all avoidable once you have seen them named.

Treating the channels as a religion

The most common mistake is the one we opened with: deciding that Amazon is "bad" or that DTC is "the future" and committing on principle. Channels are tools. A founder who refuses to use the marketplace because it "takes too much" often burns far more cash trying to buy demand for a product that had no business being on an owned site yet. The opposite founder, who treats the marketplace as the whole strategy, wakes up one day with no customer relationships and no defense against a fee increase. Hold both channels loosely and let the economics decide.

Funding DTC before the unit economics are ready

Plenty of brands launch a beautiful site, pour money into ads, and watch the cash disappear because the product simply cannot carry the acquisition cost. A site does not lower your CAC. Content, search, and brand do, and those take time to build. Launching an owned channel before you can answer the LTV-to-CAC question is the single most expensive mistake on this list.

Confusing a cheap store with a cheap channel

The software is cheap. The channel is not. We have watched founders celebrate how little their store cost to build, then quietly bleed five figures a month on ads that never reach payback. The cost was never the store. It was always the traffic, and the traffic is the part you have to plan for.

Ignoring the conversion rate of the site itself

A weak site silently taxes every dollar you spend upstream. If your store converts at one percent when a good one converts at three, you are paying three times the real acquisition cost for the same customer, and no amount of ad optimization fixes a leaky destination. Before scaling spend, make sure the place you are sending traffic is actually built to turn visitors into customers. This is the unglamorous foundation work that pays for itself many times over.

Forgetting that trust is the gating factor for unfamiliar brands

A U.S. shopper who has never heard of you needs a reason to believe you before they hand over a card. On the marketplace, the platform lends you its trust. On your own site, you have to build your own, through reviews, press, recognizable design, and a credible presence in search and in AI answers. Brands that skip this find their conversion rate stuck no matter how good the product is. Earned trust is not a nice-to-have for exporters; it is the thing that makes the whole owned channel work. We go deeper on this in digital PR as a GEO moat.

Metrics to watch: a working dashboard

If you take one operational habit from this guide, make it this: watch a small set of numbers every week and let them, not your instincts, tell you when to push and when to pause. Here are the metrics that matter, what they tell you, and a rough range to react to. As always, treat the ranges as illustrative starting points, not universal truths.

The single most important trend on this list is the share of non-paid traffic. If it is rising over time, your blended acquisition cost is falling and your channel is compounding into an asset. If it is flat and you are growing only by spending more, you have built an expensive habit, not a durable business. That one line tells you whether the owned channel is actually working the way it is supposed to.

A note for B2B exporters

Everything above assumes you sell to consumers. If you sell to businesses, the channel logic shifts in important ways, and it is worth saying explicitly because so many Chinese exporters live in the B2B world of trade shows, sourcing platforms, and long sales cycles.

For B2B, the "marketplace versus owned site" question becomes "sourcing platform versus owned brand presence." The sourcing platforms (the B2B equivalent of Amazon) hand you inbound inquiries but commoditize you against thousands of identical suppliers, and they put a middleman between you and the buyer. An owned brand presence, a credible site, real content, and visibility in search and AI answers does for B2B what it does for B2C: it lets a serious buyer find you, trust you, and reach you directly. The economics are different because the deals are larger and rarer, but the underlying principle holds. Demand you control is worth more than demand you rent.

One thing that does not change is how outreach should work, and it is worth being precise. When we support B2B clients, we research and verify a targeted prospect list matched to your ideal customer profile and deliver it to you, with contact details and a free outreach template, so your own sales team can run the conversations. We do not contact your prospects or customers on your behalf. The list and the relationships stay yours. We go deep on this model in the China B2B export playbook, on the broader B2B engine in the B2B growth engine, and on running named-account outreach in B2B ABM for named accounts.

Where free traffic comes from in 2026

The whole owned-channel argument rests on one assumption: that you can earn traffic you do not have to buy every single time. That is harder than it sounds and more important than ever, so it deserves its own section. The brands that win on their own sites are not the ones with the biggest ad budgets. They are the ones with the most traffic they did not pay for on a per-click basis.

Search and AI answers are the new front door

A U.S. customer researching a purchase increasingly starts in two places: a search engine and an AI answer engine. They type a question into Google, or they ask ChatGPT, Gemini, or Perplexity to recommend a product. If your site is structured well, your content answers the real questions, and your brand is cited by credible sources, you can be the answer they get, without paying for the click. This is the discipline we call GEO, generative engine optimization, and it is becoming as important as classic SEO. We unpack the difference in GEO vs SEO in 2026 and the specific tactics in how to get cited by AI and Google AI Overviews and traffic.

Earned media builds the trust that converts

For an unfamiliar brand, being mentioned by a credible third party does two jobs at once. It sends real traffic, and it builds the trust that makes that traffic convert. A feature in a publication a U.S. customer recognizes, a creator they follow vouching for your product, a citation in an AI answer, each of these lowers the wall of skepticism a Chinese brand faces in an unfamiliar market. This is why we treat digital PR not as a vanity exercise but as a core part of lowering blended acquisition cost over time.

The compounding effect

Here is why this matters so much for the math. Paid traffic resets to zero the moment you stop paying. Earned traffic accumulates. An article that ranks, a page that AI engines cite, a piece of coverage that keeps sending visitors, these keep working for months or years after the effort that created them. Every unit of earned traffic is a unit you do not have to buy in the rising ad auction, which means your blended acquisition cost falls even as paid prices climb. That is the entire mechanism by which an owned channel turns from an expense into a compounding asset, and it is the part that the cheap-store crowd never budgets for.

A worked scenario, start to finish

Let us put the whole framework on one brand to make it concrete. Imagine a Chinese company with a well-made $60 skincare product, a 65 percent gross margin, and early signs that customers reorder. They are deciding how to enter the U.S. Here is how the staged relay plays out.

In the first few months, they launch on the marketplace. They are not trying to build a brand yet; they are buying certainty. They learn which variant sells, what price the market accepts, and what objections show up in reviews. The fees feel high, but they are the cost of honest, fast feedback, and the cash flow funds the next step. In parallel, they stand up a clean, fast, conversion-focused site, so that any customer who searches their brand name finds a credible home rather than a dead end.

Once the product is validated, they start the slow work that actually lowers cost: content that answers the questions real customers ask, a structure that search and AI engines can cite, and a modest push for earned coverage so an unfamiliar brand starts to look trustworthy. They drop a small insert into every marketplace order inviting customers to register for a longer warranty and a discount on their next purchase, quietly moving their best customers onto a list they own. Six to twelve months in, a growing share of their traffic is no longer paid, their repeat-purchase rate is climbing, and their blended acquisition cost is falling even as ad prices rise around them. The marketplace funded the start. The owned channel became the asset. Neither would have worked as well alone.

Now imagine the same brand had skipped the validation step and poured its launch budget straight into ads for a brand-new site nobody had heard of. The CAC would have been brutal, the trust would have been missing, the conversion rate would have been low, and the cash would have run out before the channel had a chance to compound. Same product, same market, opposite outcome, decided entirely by sequence. For more teardowns of how real brands got this right and wrong, see DTC brand growth teardowns and the broader B2C growth playbook.

Frequently asked questions

If you want to know which stage your brand is actually in, and whether your unit economics can carry an owned channel yet, that is precisely what our free audit is built to answer. We map where U.S. customers are missing you, in search and in AI answers, and give you a clear read on whether the math supports building DTC now or validating further first.