How to Start and Run an E-commerce Business
Most online stores do not fail because the website was ugly. They fail because the arithmetic never worked — the product cost too much to buy, too much to ship, and too much to advertise, leaving nothing behind. Before you design a logo, it is worth understanding the shape of the business you are about to run.
Decide what you are actually selling
A good first product is boring to look at and easy to explain. You want something with genuine demand, a price high enough to absorb advertising costs, and no fragile, oversized or hazardous shipping profile. Steer clear of anything needing certification you do not hold, such as electricals, cosmetics, supplements or children's toys, until you understand the compliance burden.
Sanity-check demand before you commit money:
- Search the product on the marketplaces your buyers use. Steady listings with hundreds of reviews mean real demand; an empty category usually means no demand rather than no competition.
- Look at how many sellers already offer the identical item. If forty stores sell the same white-label product, your only lever is price, and that is a race you will lose.
- Read one-star reviews of competing products. The complaints tell you what to fix, and a fixed complaint is a reason to buy from you.
Pick a business model
Four models cover most stores, and they differ mainly in who holds the stock and the risk.
- Buy and hold stock. Best margins and full control of quality and delivery speed, but your cash is tied up in inventory that may not sell.
- Dropshipping. No stock, so almost no upfront cash, but thin margins and no control over delivery or packaging.
- Print on demand. The supplier prints and ships per order. Good for designs and merchandise, capped margins.
- Digital products. Software, templates, courses. No shipping and near-zero marginal cost, but the work is all in marketing and support.
Do the unit economics before anything else
Work out the profit on a single order, because every problem in e-commerce is visible at that level. Take your selling price, then subtract the cost of goods, inbound shipping and duty, outbound delivery, packaging, payment processing of roughly two to three per cent, marketplace or platform fees, and an allowance for returns.
What is left is your contribution per order, and it has to cover advertising. If a product sells for 40 and leaves 14 after all of the above, then a customer acquisition cost above 14 means you lose money on every sale — and you lose it faster the more you spend.
Two numbers decide whether the business compounds or bleeds:
- Customer acquisition cost (CAC) — total marketing spend divided by new customers won.
- Lifetime value (LTV) — contribution from the average customer across every order they will place.
A ratio of roughly 3:1 between lifetime value and acquisition cost gives you room to pay for overheads and still grow. If they are close to equal, you are working for the ad platform rather than yourself.
Get the boring setup right
Register the business properly for your country, keep business banking separate from personal from day one, and understand your sales tax or VAT obligations before you cross a registration threshold rather than after. Distance-selling rules in most markets give buyers a cooling-off period and a right to return goods; your returns policy must reflect the law, not just your preference.
Write your shipping, returns, privacy and terms pages before launch. Payment processors ask for them, and shoppers genuinely read the delivery and returns pages before a first purchase.
Operations are the real job
Once orders arrive, the business becomes logistics and communication. Decide who packs orders and how quickly, what your dispatch cut-off is, and what happens when something arrives broken. Set delivery expectations you can beat rather than ones you hope to meet — the single largest driver of complaints and chargebacks is a delivery date that slipped.
Answer customer messages within one business day. Support is not overhead; a well-handled problem produces more repeat custom than a sale that went smoothly.
Measure a small number of things
- Conversion rate — visitors who buy. Under one per cent usually points at price, trust or delivery cost.
- Average order value — the cheapest number to improve, through bundles and thresholds for free delivery.
- Contribution margin per order — the truth about whether you can afford to advertise.
- Repeat purchase rate — whether you are building an asset or renting traffic.
- Refund and dispute rate — an early warning about product quality or expectations.
Where new stores go wrong
- Spending on advertising before the product page converts organic visitors.
- Competing on price against sellers with far greater buying power.
- Ignoring delivery cost until it eats the margin.
- Adding hundreds of products instead of making a handful sell well.
- Treating returns as an exception rather than a normal, budgeted cost.
Start narrow, sell a small range properly, and only widen the catalogue once the numbers on a single order genuinely work.