Selling on Amazon vs your own store: what each one is for

The question gets framed as a choice, and for most brands it is not one. Your customers are in both places, and refusing to meet them somewhere costs more than it saves.

The expensive mistake is not choosing wrong. It is running the same copy, the same images and the same argument in both, then wondering why one of them underperforms.

The marketplace is not a phase

Start here, because a lot of brand-side thinking still treats marketplaces as a compromise to be escaped.

Roughly half of US product searches begin on Amazon. Three separate 2023 surveys put it at 49%, 50% and 57% - CivicScience, PowerReviews and Jungle Scout respectively, different methods and sample sizes landing in the same place. Take the range rather than any single figure, and note they are all a few years old, but the shape is not in doubt.

In Southeast Asia the same behaviour runs through different platforms. Shopee holds somewhere around half of the region's e-commerce, with Lazada and TikTok Shop taking most of the rest, in a market that was about $159 billion in 2024 and is forecast past $200 billion. A Malaysian shopper looking for a camping stove opens Shopee the way an American opens Amazon.

This is structural, and the reason is not convenience. It is guaranteed trust. The marketplace underwrites the return, the delivery and the refund, so the shopper does not have to decide whether to trust you. They only have to decide whether they want the product. That is an enormous thing to be handed, and it is not going away.

So the honest position is not marketplace-versus-brand. It is that one channel sells to people who have not decided to trust you yet, and the other sells to people who have.

The structural difference

On your own store, you control the sequence. Someone can arrive on a blog post, read why the product is built the way it is, meet the founder, look at three products and buy on their second visit a week later. You decide what comes first, second and third. You own the pace.

On Amazon, there is no sequence. There is one screen, mostly decided before anyone reads a full sentence, sitting beside competitors, reached through a search box people use like a filing system.

That is not a worse environment. It is a different job. One is an argument you build. The other is a claim you land immediately.

The most underused asset on Amazon is the search data

Everyone talks about marketplace fees. Almost nobody talks about what you are being given in exchange, which is the only large-scale record of how customers describe what you sell in their own words.

Not what you call it. Not what the brand deck calls it. What they type when they want one.

Search term data tells you three things you cannot get anywhere else:

• The words. The actual nouns people reach for, which are frequently not yours.
• The functions. What problem they are shopping for, visible in the qualifiers they attach - waterproof, foldable, for small kitchens, gift.
• The comparison set. Who they think you are like.

Your own analytics cannot do this. They show what visitors did after they found you. They are silent on the language of everyone who never did.

Always run an auto campaign

This is the single most practical thing in this piece, and it is the thing I see skipped most often.

Run an always-on automatic campaign. Not because it is the most efficient spend - it usually is not - but because nothing reads your listing more literally than Amazon's own matching algorithm. You are paying a small tuition fee for an unsentimental machine to tell you what it thinks you sell.

Then read the search term report and ask two questions:

Are you showing up for the right keywords?

Are you showing up against the right competitors?

If the answer to either is no, you have not found an advertising problem. You have found a language gap - the distance between what you mean and what the market has understood. The algorithm is simply the fastest and cheapest instrument for measuring it, because it has no opinion about your brand and no reason to be polite.

I find this more useful than most brand research, and it costs a fraction. A customer in a focus group will tell you what they think you want to hear. A machine matching your listing to searches will show you exactly which shelf it has filed you on. Brand love, measured in customer language, is the same test run on reviews instead of searches.

What Amazon cannot do

It cannot give you the customer. You get an order, not an email address, a relationship or permission to speak again. Every sale is a fresh acquisition.

It cannot carry a long argument. Anything that needs context or trust built over time does not fit on one screen.

It cannot protect you from comparison. Your product sits next to a cheaper one, permanently.

It cannot be changed on your schedule. Character limits, formatting rules and policy shift without warning, and your listing complies or it stops ranking.

That last one is not theoretical. When we rebuilt Rohit's Mithai's Amazon listing, part of what we found was a title breaking Amazon's new length limit and bullets running past the character cap. Organic visibility had fallen 51% in six months while demand held steady. Nothing about the product had changed. The rules had.

What we found at Rohit's

Rohit's Mithai makes burfi - the Indian sweet passed around at weddings and at Diwali - remade to keep the texture without the sugar load. It sells in the US through its own store and through Amazon.

The brand had everything a brand is supposed to have. A mission, a story, an identity, a founder who lives it. Shoppers on Amazon were getting none of it. They saw a premium price on a familiar category and stopped there.

That is a positioning problem, not a price problem. Rohit's is not a more expensive mithai. It is a traditional sweet rebuilt into something you can eat without guilt and still give as a gift, which is closer to a new category than a better version of an old one.

So the listing had to carry the whole argument - a healthy treat that is also a gift, in the title, the images and the first two bullets, in the words shoppers actually type. We started from search query analysis rather than the brand's own vocabulary, rewrote the title, bullets and backend terms inside Amazon's current limits, built an image stack and A+ content to continue the story where the copy runs out, and restructured Sponsored Products to stop paying for searches that never converted.

Listing conversion went from 1.99% to 4.62%. Sales rose 505% from July to August. Advertising cost of sale fell from 35.9% to 21.1%. All three are published on the case study with the client's approval.

The point is not the lift. It is that the product, the price and the brand strategy did not change. What changed was the translation of an existing position into a format that gives you one screen.

Your own store is where brand love lives

None of this makes the webstore optional. It makes it a different job.

Amazon can sell your product. It cannot let anyone step inside your brand. There is no room on a listing for the founder, the method, the reason for the price, the world the product belongs to - and no permission to build any of it over time, because you will never speak to that buyer again unless they choose to find you.

Your own store is the only place a customer can stop being a shopper and start being someone who belongs to something. That is where loyalty is built, and it is the only place brand love can form - because brand love is a customer being able to describe you, and nobody learns to describe a brand from a bullet list.

The best of both is a one-way transfer. Take the unglamorous data Amazon hands you - the words, the functions, the comparison set - and apply it deliberately on your own store, where you have room to do something with it. Amazon tells you what people call it. Your store is where you explain why yours is different.

Making your own store earn more

Everyone wants this, because marketplace fees genuinely hurt and every brand would rather sell direct.

There is no shortcut, and I want to be fair about why. Amazon's teams have done a stellar job. The checkout, the search, the delivery promise, the returns flow - that experience is the product of enormous sustained engineering, and a shopper who has been trained by it arrives at your store with that as the baseline. You are not competing with a website. You are competing with a standard.

But there is one thing their developers and commercial teams cannot build for you, and it is the only thing that will ever pull someone off a marketplace: your brand world. Nobody can engineer that on your behalf.

So the work is consistency, and it is testable:

• Test the language. Not just the headline. The button, the confirmation, the shipping note.
• Test the visuals. Does the photography say the same thing the price does?
• Test the promotions. Same discipline as anywhere else: a reason before a number.
• Test the UX, and take it seriously. This is the one brands skip because it feels like a developer's job.

The question to hold over all of it: how easy, and how delightful, is it to shop here? Easy is the engineering. Delightful is yours.

Shopping is anticipation, not acquisition

Here is the part that most direct-to-consumer stores design past.

Buying something is a small, real pleasure, and most of that pleasure is not in owning the thing. It is in the anticipation - the moment the order goes through, and then the days of waiting for the parcel. The purchase is a promise of something arriving.

A marketplace is efficient at removing friction from that moment. It is not especially interested in extending it, because it has ten thousand other brands to ship.

You can be. The order confirmation, the shipping note, the message in the box, the email that lands the morning it is out for delivery - each of those is a chance to keep the anticipation alive rather than process it. Most brands treat them as system receipts and let a platform default write them.

Make them delightful. Make them warm. A transaction ends at the payment. A relationship runs through the wait. That gap is the whole opportunity, and it costs almost nothing except attention.

Which to start with

Short on demand, long on story: Amazon first. Buyers with intent, borrowed trust, and the search language to sharpen your positioning. Bring all three back to your own store.

Short on margin, long on traffic: your own store first. Fees and ad costs are a real tax, and if you already have an audience you are paying a toll to reach people you could reach directly.

Short on both: your own store, and a narrower range. Amazon rewards brands that already know what they are. Going there to find out is an expensive way to learn something a fortnight of customer conversations would tell you.

There is a great deal more in the distinction between a marketplace listing and a brand's own store - the fee arithmetic, what actually moves a marketplace buyer onto your site, and how far the two prices can diverge before one of them looks wrong. That is a piece of its own, and it is coming.

For now: open your Amazon search term report and your homepage side by side. Are they describing the same product?