Every omni-channel expansion plan we review covers the same ground: which channel, what margin, what the fee structure looks like, how inventory gets there, who owns the listings. Good plans go further and cover pricing architecture and channel sequencing.
We have never once been handed an expansion plan with a returns section in it.
That’s not a criticism of anyone’s planning. It’s a symptom of where these brands learned logistics. If your operational instincts were formed inside FBA, returns are not a process you run — they’re a thing that happens to a number in a report. A customer sends it back, Amazon receives it, grades it, restocks or disposes of it, and the only artefact that reaches you is a line in a reimbursement file and a rate you check quarterly. Nobody in your business touches a returned unit. Nobody has ever needed to.
Then you launch on Walmart, TikTok Shop and your own Shopify store, and three channels’ worth of product starts coming back under three different rule sets to an address nobody has decided on yet.
The structural fact almost nobody checks first
Here’s the one that catches the most brands, and it catches the sophisticated ones specifically.
A lot of brands solve multi-channel fulfilment with Amazon MCF — you’re already holding inventory in Amazon’s network, so you draw on the same stock to pick, pack and ship orders placed on other channels. It’s a genuinely good answer to the outbound problem, and Walmart now permits it provided you use unbranded packaging, block Amazon Logistics as a carrier and meet the shipping windows.
But per Amazon’s own Supply Chain documentation, MCF Shipping does not manage returns for you.
That sentence does more damage than any fee schedule in this post. A brand that solved outbound with MCF very reasonably concludes it has solved multi-channel logistics. It hasn’t. It has solved one direction. Product flows out of Amazon’s network to a TikTok Shop customer, and when that customer sends it back, there is no equivalent path in reverse. You either run your own returns process to your own address, or you direct the customer to ship it back into your Amazon inventory — which raises its own questions about grading, condition and whether you actually want a channel-agnostic customer return landing back in FBA sellable stock.
The failure mode isn’t dramatic. It’s that nobody made a decision, so the answer defaults to whichever address the customer service inbox improvised the first time it came up.
The three channels do not agree on who pays
The second thing that breaks is the assumption that returns economics travel with the product. They don’t. Each channel has its own settlement of the same question — who eats the return shipping and who eats the unit — and the answers are materially different.
Amazon. Since February 2026, sellers bear return shipping costs on seller-fulfilled orders via prepaid labels. On FBA, returns are handled inside the network and you absorb the unit outcome — restocked, unsellable, disposed — plus fees in both directions.
Walmart. The answer depends entirely on fulfilment method, and this catches brands who assume Walmart works like Amazon because the marketplace looks like Amazon. Under Walmart Fulfillment Services, Walmart handles returns and post-sale support in a model that broadly mirrors FBA. Under standard Marketplace fulfilment, you own it end to end — processing, return shipping, customer service, the lot. Sellers set their own return rules inside Walmart’s framework, which means the return policy on your Walmart listings is a decision somebody in your business made, or failed to make, and probably can’t currently recite.
TikTok Shop. Different framework again. Reporting on the January 2026 policy update describes a 50/50 cost split on non-seller-fault returns and return shipping subsidies tiered by seller performance score — with sellers at the top of the score band receiving partial coverage and sellers below the threshold receiving none. Treat the specific percentages as directional and confirm the current terms in your own seller centre, but the shape is not in dispute: your return economics on TikTok Shop are a function of a performance metric, which is a mechanic that exists nowhere in your Amazon operation.
Three channels. Three different answers to “who pays.” One returns process, designed for the channel that answers it for you.
The volume assumption is wrong too
The economics are one problem. The rate is a separate one.
Industry reporting consistently puts TikTok Shop return rates well above Amazon’s — commonly cited in the 8-15% range against Amazon’s low-to-mid single digits, with apparel and fashion at the top of the band and home goods at the bottom. Treat those as directional rather than a benchmark you plan against; category variance is enormous and nobody publishing these numbers has your catalog.
But the direction is well-supported and the mechanism is obvious to anyone who has watched the two buying behaviours. Amazon is search-first: the shopper arrived with intent, compared options, read reviews, and made a considered decision. TikTok Shop is discovery-first and frequently impulse-driven: the shopper was not looking for your product ninety seconds ago. A purchase decision made in under a minute produces a different return rate than one made over three days, and no amount of fulfilment competence changes that.
So the brand that models a new channel at Amazon’s return rate is modelling the wrong denominator, and the error compounds: return rate feeds directly into contribution, and every few points of return rate takes real percentage points out of net margin on a mid-priced physical product.
Why this surfaces as a margin mystery instead of a returns problem
The reason this is worth writing about rather than just fixing quietly is the diagnosis pattern, which we see constantly.
Returns are recorded per channel, but the costs of a broken returns architecture leak into lines that aren’t channel-tagged. Return shipping you didn’t budget for shows up in freight. Units that came back to a warehouse with no grading process show up as inventory shrink or as unsellable stock somebody eventually writes off. Customer service hours spent improvising a return path for a channel with no documented policy show up in payroll. Time spent by an ops lead untangling a TikTok return that arrived at the 3PL with no RMA shows up nowhere at all.
Six weeks later, blended contribution is down two points, and the conversation in the room is about Amazon ad efficiency — because advertising is where the reporting lives and where somebody is already looking. Nobody’s first instinct is that the margin leak is in reverse logistics on a channel that represents 9% of revenue.
We have watched brands restructure campaigns over a variance that originated in a returns process nobody had written down.
The timing problem is specific and it is now
Two dates make this an August conversation rather than a someday conversation.
Expansion decisions are being made this month. Brands finishing mid-year reviews and deciding to be on three channels for Q4 are deciding it right now, because inventory commitments and channel onboarding both need lead time before peak.
The returns wave lands in January. Holiday purchases return in the new year. For a brand that expanded in Q3, this coming January is the first January in which returns arrive from three different channels under three different rule sets simultaneously — during the exact weeks your team is reconciling the year, your ops lead is taking deferred time off, and your cash position is at its most sensitive.
That is not a good week to be designing a process. It is an excellent week to be executing one you wrote in August.
The architecture, in the order it has to be decided
Six decisions. None of them require software. All of them require somebody’s name attached.
1. Name the physical destination for each channel, in writing. Where does a returned unit from Walmart physically go? From TikTok Shop? From DTC? If the answer for any channel is “I’d have to check,” that is the finding. Consolidating to one address is usually correct and is not automatic — MCF won’t do it for you.
2. Write the return policy per channel and know where it differs. On Walmart Marketplace you set your own rules within their framework. On TikTok Shop the platform’s framework does more of the work. Assign one owner who can produce all three policies on request. If your policies differ across channels for no deliberate reason, they were set by whoever configured each account on the day they configured it.
3. Decide the grading and disposition process before you need one. FBA graded your returns for you. Nobody else does. Who inspects a returned unit, against what standard, and what happens to sellable, refurbishable and dead stock? A brand without this answer defaults to a pallet in the corner of a 3PL that becomes a write-off in Q2.
4. Model each channel on its own return rate. Not Amazon’s. Build the contribution model for a new channel with a return rate assumption that reflects that channel’s buying behaviour, and be explicit that it’s an assumption until you have ninety days of your own data.
5. Budget the return shipping line per channel. Who pays is a different answer on each platform, and on at least one of them it moves with your performance score. That is a real, forecastable cost line that appears in almost no expansion model we’re handed.
6. Baseline before you launch, not after. Pull your current Amazon return rate by ASIN and your current cost per return before the new channel goes live. You cannot detect a multi-channel returns problem without a single-channel baseline, and almost nobody pulls one.
The part that’s actually good news
Reverse logistics is unglamorous, which means it’s under-competed. Most of your competitors expanding into the same channels this quarter are making exactly the same omission, and they will find out in January along with everyone else.
There’s also a genuine strategic asset buried in it: a return is the most expensive feedback your listing can produce, and it’s the only feedback from someone who already bought. Once you have a real returns process on a second channel, the reason codes and buyer comments from that channel tell you something your Amazon data can’t — whether the expectation gap is in your creative or in the buying context. A product that returns at 4% on Amazon and 12% on TikTok Shop is not a worse product on TikTok. It’s a product whose page did the persuading on Amazon and whose fifteen-second video did it on TikTok, and only one of those set an accurate expectation.
That’s a creative finding you can only get by expanding — and only if somebody is capturing the returns instead of absorbing them.
FAQ
Does Amazon MCF handle returns for my Shopify or TikTok orders?
No. Per Amazon’s Supply Chain documentation, MCF Shipping does not manage returns for you. You either run your own returns process to your own address or direct the customer to ship back into your Amazon inventory, and both are decisions with consequences you should make deliberately rather than discover.
Should I just use one return policy across every channel?
Usually yes for the customer-facing terms, because a single policy is easier to staff and easier to defend, and inconsistency across channels is a support burden. But the cost structure underneath will still differ by platform, so a single policy does not mean a single cost model.
Can I route all multi-channel returns back into FBA sellable stock?
Physically, in some configurations. Commercially, think hard about it. You are putting a unit that was graded by nobody into inventory that Amazon will sell as new. If the disposition process isn’t real, this is how a return rate on one channel becomes a review problem on another.
How much return rate difference should I actually plan for on a new channel?
Plan on more than Amazon and refuse to be precise until you have your own ninety days. Reported ranges put TikTok Shop meaningfully above Amazon with heavy category variance. Anyone handing you a confident single number for your specific catalog is guessing.
We’re already live on three channels and none of this is documented. Where do we start?
Question one. Find out where units are physically arriving for each channel today, and how many arrived last quarter. That single answer usually surfaces the entire problem in an afternoon, and it frequently surfaces stock nobody knew was sitting somewhere.
—
If you’re looking for a team that manages every lever — creative, advertising, and operations — Velocity Sellers works with brands doing $100K+/month on Amazon. Contact us for a free account audit.
