A catalogue with a Buy button is not a store. A store begins where the order reaches the warehouse, the payment reconciles with the bank, and the product page stops showing "in stock" for something that sold yesterday. Below is every moving part in order: platform choice, payments and shipping, the product page, checkout, and what breaks in month two — with the numbers you need to plan a budget.

Platform: three scenarios, not thirty

Comparing twenty content management systems wastes a week. The real choice collapses into three scenarios, plus a zeroth one — validate demand on a marketplace first. The deciding question is not "which platform is best" but "how many SKUs do I have and how unusual is my pricing logic".

Payments and shipping: the underestimated node

This is the most common reason a launch date slips — not because the integration is hard, but because it depends on third parties and paperwork. Allow two to three calendar weeks even though the work itself is about five days.

• Card acquiring — 1.8–2.8% per transaction. Approval takes 3 to 10 working days and requires a registered business entity, bank details, and a live site with published terms.
• Cash on delivery — still 40–60% of orders in many categories. You need partial-prepayment logic, otherwise your collection rate drops.
• Courier and parcel-locker integration — not "add an address field", but a live API for locations with autocomplete plus label generation from the admin. That saves 2–4 minutes per order.
• Apple Pay and Google Pay — the cheapest available lift in mobile conversion, typically 5–12% more completed orders, because manual card entry disappears.
• Instalments or buy-now-pay-later — worth it above an average order value of roughly $150. Below that the integration never repays the 15–25 hours it takes.

The product page: what actually moves a decision

The product page absorbs most of your search traffic and is where the decision happens. On projects where we rebuilt only that page and touched nothing else, conversion rose 15–30%. The cause is always the same: the buyer was missing facts, not persuasion.

On copy: a 200-word description paraphrasing the product name helps neither the buyer nor search. Write against the questions people actually type — "will it fit", "how do I clean it", "how is it different from". Gathering those queries is covered in our guide to SEO for online stores.

Cart and checkout: where the money leaks

Seventy percent cart abandonment is not a disaster, it is the market baseline. But roughly a quarter of those carts are recoverable through checkout mechanics alone, without discounting anything.

1. 1

   The cart is a panel, not a separate page

   Adding an item should not eject the shopper from the catalogue. A slide-over panel with a "keep shopping" button typically adds 10–20% more line items to the average order.

2. 2

   Guest checkout is the default

   Mandatory registration costs about 20% of orders. Create the account silently after payment and email credentials later, if an account is needed at all.

3. 3

   One screen, 5–7 fields

   Name, phone, shipping method, pickup point, payment method. Drop middle names, postcodes and order comments — the postcode comes from the lookup and the comment matters on one order in twenty.

4. 4

   Full total before the card form

   The single most common abandonment trigger is shipping cost appearing on the last step. Show the final number before you ask for money.

5. 5

   Errors next to the field, never at the top

   Validate on blur and put the message under the input. A red banner at the top of the page frequently sits off-screen on a phone.

After launch, stop trusting intuition: wire cart and checkout events into analytics and watch which step people vanish on. The method for running those checks in sequence is in our guide to conversion rate optimisation.

What breaks in month two

The store you sign off and the store six weeks later are different systems. Here is the order in which problems usually appear, and what to build in ahead of time.

1. 1Stock levels drift from reality. Manual updating survives up to about 20 orders a week; past that you need inventory sync every 15–30 minutes.
2. 2Orders get lost between email and messenger threads. A CRM integration solves this: every order becomes a record with a status and an owner.
3. 3Transactional emails land in spam. Configure SPF, DKIM and DMARC at launch rather than after the complaints — a day of work.
4. 4Filtered catalogues slow down past 500 products, nearly always for the same reason: filtering pulls the full product list into memory instead of querying an index.
5. 5Returns and partial cancellations were never designed into the admin. Work that scenario out before launch, or finance will run it in a spreadsheet.

Frequently asked questions about building an online store

In short

• Choose a platform by SKU count and pricing complexity, not by fashion. Under 200 products a hosted builder usually wins on total cost.
• Payments and shipping are a calendar risk, not a technical one. File the acquiring paperwork in week one.
• The product page takes most of the traffic: images, stock status, shipping cost and specifications matter more than elegant copy.
• Guest checkout on a single screen with the full total shown before the card form recovers part of that 70% abandonment.
• Plan inventory sync and CRM before launch — manual handling collapses at around 20 orders a week.