Start with a customer, not a storefront.

1. Define the buyer and the job

Write one sentence: “I help [specific buyer] get [outcome] when [situation].” Then interview five people who match that description. Ask what they last bought, what disappointed them, what they compared, and how they decided. Avoid asking whether they would buy your hypothetical product; past behavior is stronger evidence than a polite prediction.

Record the alternatives people already use. Your competitor may be another store, a local service, a spreadsheet, or doing nothing. If you cannot name the existing alternative, continue researching before buying stock or paying for ads.

2. Calculate the real cost per order

List product cost, inbound freight, packaging, payment fees, outbound shipping subsidy, expected returns, and customer acquisition. Compare the total with the selling price before assuming a margin. For example, a €40 sale with €15 product cost, €4 fulfilment, €2 payment fees, €3 expected returns and €8 acquisition leaves €8 before fixed costs and tax. If acquisition rises to €13, the contribution falls to €3.

Use your own supplier quotes and current fees. Model a base case and a worse case: slower delivery, more returns, or higher ad costs. Do not treat revenue as profit.

3. Verify supply and delivery

Request a sample, confirm the product specification, and measure the actual delivery time. Ask the supplier who handles defects and whether stock information is reliable. If using dropshipping, you still own the customer experience: the buyer needs accurate delivery estimates, clear support, and a workable refund process. Put a backup supplier or a pause rule in writing.

4. Build one trustworthy buying path

Start with one product page. Show accurate photos, dimensions or compatibility, the full price, shipping estimate, returns terms, and a reachable support contact. On mobile, a buyer should be able to understand the offer and reach checkout without hunting through menus. Test address entry, failed payment, confirmation email, and an actual refund path using your payment provider’s approved test flow.

5. Run a bounded launch experiment

Choose one traffic source and a fixed test budget. Track visitors, product-page views, add-to-cart actions, completed orders, acquisition cost, refunds, and contribution per order. Decide in advance what would make you continue, change the offer, or stop. Ten orders can reveal operational problems; they cannot prove long-term demand.