Your e-commerce business can be growing while your margins move in the wrong direction.
U.S. retail e-commerce sales reached $340.2 billion in the second quarter of 2026, up 12.2% from a year earlier, according to the U.S. Census Bureau. Online demand is still growing. For a reseller, though, more demand doesn’t solve the basic problem. If ten other sellers can buy the same product, you have fewer ways to stand apart.
You can add more SKUs. You can sell through Amazon, Walmart Marketplace, eBay, Etsy, TikTok Shop or your own store. Those moves can expand your reach, but you’re still competing with products other sellers may also carry.
Owning the product gives you another option. You control what gets made, how it’s designed and why a customer should choose it.
Why Owning the Product Changes the Business
Reselling works. Plenty of good e-commerce businesses are built that way. The problem appears when your product becomes interchangeable with everyone else’s.
If another seller carries the exact same item, customers often compare price, delivery speed, reviews and listing quality. Cutting the price might increase sales, but every discount comes out of your margin.
Developing your own branded product gives you more control over the reasons someone buys from you. You can change the materials, solve a complaint that shows up repeatedly in reviews, improve the packaging or build a feature for a specific customer.
You also don’t have to invent something the world has never seen. Some good product opportunities start with an existing category and a clear problem customers still complain about.
That’s an important distinction. A private-label product, an improved version of an existing product and a completely new product require very different amounts of research, design work, engineering and capital.
At Gembah, we treat those early decisions as part of the product itself. What you decide during research affects the design. The design affects factory options. Factory options affect unit cost, tooling and lead time.
Making those decisions separately can get expensive.
Decide What Kind of Product You’re Building
Before spending money on CAD files, samples or tooling, define what you’re actually trying to create.
You may have found an existing product with weak reviews and an obvious opportunity to improve it. You may already sell in a category and see something your customers keep asking for. Or you may have a new product idea that requires original design and engineering.
Those aren’t the same project.
A modest change to an existing product might require research, design changes, sourcing and sampling. A new mechanical or electronic product may need industrial design, engineering, several prototype rounds, testing and custom tooling before a factory can produce it.
The more original the product, the more questions you need to answer before production.
That’s where we’d rather slow a project down. Paying for another design revision is usually easier than discovering the problem after you’ve paid for tooling or placed a large purchase order.
Money: Work Backward From the Business Model
Your budget matters, but the more useful number is what the product can afford to cost.
Start with a realistic selling price. Then account for marketplace fees, fulfillment, freight, duties, packaging, returns, advertising and the margin your business needs.
What’s left gives you a much better idea of what the product can cost to manufacture.
If you expect to sell something for $40, falling in love with a design that costs $28 to make creates a problem. Producing more units won’t fix weak unit economics.
This is one of the first places where product research and product design need to work together. The feature customers love may also be the feature that requires an expensive material or complicated assembly. You need to know that before it becomes part of the final design.
Your development budget may also need to cover industrial design, engineering, prototypes, intellectual property work, tooling, samples, packaging and testing. Some products require certification or compliance work as well.
Don’t commit most of your available cash to the production order and assume everything before that will be cheap.
Good development work costs money. So do mistakes. The difference is that good development spending should answer questions before the expensive commitments begin.
People: Bring in Expertise at the Right Stage
You can lead a product project yourself. You probably shouldn’t expect to do every part of it yourself.
A physical product may involve market researchers, industrial designers, mechanical or electrical engineers, prototype specialists, sourcing professionals and factory quality teams.
You won’t need all of them on every project.
We’d rather bring in the right expertise when the product calls for it than build a large team around a simple project. Category experience matters too. Someone who understands textile construction brings different knowledge than an engineer working on a connected electronic device.
The same applies to manufacturing. A factory that does excellent injection molding may be the wrong choice for a product that depends on textiles, electronics or specialized assembly.
The job is to match the people and production capabilities to the product you’re actually building.
If You Want to Improve or Change an Existing Product
Existing products give you something valuable to study. Customers have already told you what they like, what breaks and what they wish worked differently.
Start there.
Marketplace search data can help you measure demand. Reviews and return complaints can show where current products disappoint buyers. Competitor pricing helps you understand how much room you have to improve the product without pricing yourself out of the category.
That research should lead to a specific reason for creating another version.
“We can sell this too” isn’t much of a product strategy.
A better reason might be that customers repeatedly complain about a weak hinge, awkward storage, poor fit or packaging that causes damage. Now your design team has a problem worth solving.
You also need to check intellectual property before copying product features or mechanisms. An existing market proves that customers buy the category. It doesn’t give you permission to reproduce protected designs or technology.
Once you define the change, designers can turn it into sketches, CAD models and production specifications. This is also where manufacturing needs to enter the discussion.
A design can look good on a screen and still be expensive to produce. A small feature may require another mold, another assembly step or a material that only a few factories can handle.
We’d rather resolve that in the design files than after tooling begins.
Factory selection deserves the same discipline. Don’t compare quotes until suppliers are quoting the same specifications, materials, finishes and quality requirements. A cheaper quote doesn’t help if the factory interpreted the product differently.
Where you manufacture also depends on the product. China remains a major manufacturing base, while Vietnam, Mexico, the USA and other markets can make sense based on the production process, order volume, lead time and supply requirements.
If you’re considering manufacturing in China, local supplier knowledge and factory oversight can matter as much as the initial quote. A supplier profile and a few emails don’t tell you enough about how a factory will handle production.
If You Want to Create a Completely New Product
A new product starts with a different set of questions.
Define the customer problem first. Then decide which features the first version needs to solve it.
This is where founders often make development harder than it needs to be. Every feature adds something. It may add parts, tooling, engineering work, testing, assembly time or another way for the product to fail.
More features don’t automatically make a stronger first product.
Early sketches are useful because they’re cheap to change. CAD models take the idea further. Engineering answers questions about how parts fit, move, carry loads, manage heat or connect electronically.
Then prototyping lets you test the assumptions behind the design.
Your first prototype doesn’t need to look ready for a retail shelf. It needs to answer useful questions.
Does the mechanism work? Is the size right? Can someone use it the way you expected? Does the material feel wrong? Is an important feature creating more trouble than value?
Each prototype round should have a reason.
Validation also needs better input than asking friends whether they like your idea. Friends tend to be supportive. Customers spending money behave differently.
Depending on the product, you can use customer interviews, landing-page tests, small paid campaigns, preorders or crowdfunding platforms such as Kickstarter and Indiegogo to test interest.
Once the design is stable enough for manufacturing, the factory needs clear specifications. Dimensions, materials, tolerances, finishes, performance requirements and packaging expectations should be documented before you compare final production quotes.
Alibaba can help you find supplier candidates. Treat it as a starting point, not a factory approval process.
Before placing an order, confirm what the supplier actually manufactures, what it subcontracts and whether it has experience with products similar to yours. Review samples against your specifications. Determine how quality inspections will work before production begins.
Factory price is one part of the decision. Quality, communication, production capability and consistency matter once real inventory is on the line.
Technology: Use Better Data, but Don’t Mistake It for Proof
Product research has changed a lot since this article first appeared.
In 2026, sellers have faster access to marketplace data, keyword demand, competitive pricing and customer review patterns. AI tools can also help sort large amounts of feedback and identify recurring complaints faster.
Tools such as Helium 10, ZonGuru and Jungle Scout can help Amazon sellers compare products, categories and demand signals.
The tools are better. The basic discipline hasn’t changed.
A high search volume doesn’t prove you should build a product. Estimated sales don’t tell you why customers choose one item over another. AI can group complaints, but somebody still needs to decide which complaints represent a worthwhile product opportunity.
Use the data to narrow the field. Then validate the customer, product and economics before spending heavily on development.
The same discipline applies once development starts. Keep current specifications, design revisions, factory quotes, sample feedback and quality requirements organized.
A supplier manufacturing from an old drawing can create a very expensive misunderstanding.
At Gembah, research, product development, design, engineering, sourcing and manufacturing sit within the same process. That connection matters because a decision made during design can change your production cost months later.
Know What You’re Committing To Before Production
Your first branded product doesn’t need every possible feature. It needs a clear customer, a good reason to exist and economics that work.
It also needs to be something a factory can make consistently at the quality and cost your business requires.
That means doing some work before the exciting part. Study the market. Define the product. Test the risky assumptions. Understand the production cost. Get the specifications right before asking a factory to commit.
You can improve the product after launch. Changing a sketch, CAD file or prototype is part of development. Changing finished tooling or sitting on thousands of units you can’t sell is a much harder lesson.
If you’re considering moving from reselling into your own branded product, talk through the idea before committing to tooling or a production order. Gembah can help you assess the opportunity, define the product and work through design, prototyping, sourcing and manufacturing with the later production decisions in mind.
Gembah can help you pressure-test the idea, work through design and prototyping, and prepare for sourcing and manufacturing before expensive decisions get locked in.
