Shopify store setup

Why owners come to this page

Two different people land here and only one of them ever says the word Shopify.

The first is already selling and it is working well enough to hurt. Orders arrive at eleven at night and are buried under new messages by morning. Two customers are sold the same last item because the stock count lives in someone's head. Somebody who has already paid asks three times whether the money arrived. An address gets written down half-finished and the parcel comes back. And when the owner travels for two days, the shop shuts, because the shop is the owner. That is not an inconvenience. That is a business being crushed by its own success.

The second already has a store and it is not carrying its weight — a theme that never got finished, a catalogue half-loaded, payments that half work, and a checkout nobody has ever completed on a phone.

Either way, what is actually being bought is not a website. It is a machine that takes money and keeps count of stock without a person standing behind it. Four things have to be joined and stay joined. The goods — products, images, variants, how many are left. The money — which gateway, whose account it lands in, how a refund travels back. The delivery — which zones, what it costs, how many days, what happens when it comes back. The customer — who they were, what they bought last time, how they are brought back. Everything else is detail hanging off those four.

The difference between this and ordinary web work is what a mistake costs. On a brochure site, a small mistake means the page looks wrong. Here it means a card charged twice, or an item sold that is not in the warehouse, or a delivery rate set slightly wrong that loses money on every order for three weeks before anybody notices.

How the build runs

The order below is deliberate. Product data comes before design, because product data is what actually holds a launch date hostage.

1. Discovery and a scope with counts in it. Not "load the products" — twenty-eight products, up to three variants each, using images and copy you supply. Numbers in a scope end arguments before they start. We also agree here what is not included, what you owe us and by when, and how many rounds of design feedback the price covers. Two to three days.
2. Catalogue as a spreadsheet, approved before it is imported. Every product, variant, SKU, price and stock level goes into a CSV first. You read the CSV and approve it. Then it is imported in one pass. If prices change in month two, the CSV changes and re-imports — nobody opens twenty-eight admin pages by hand. Three to five days, more if the data is arriving in pieces.
3. Theme, navigation and brand. Dawn or a paid theme, configured properly rather than hacked; main menu, footer, collection structure, policy pages. Custom code only where a theme setting genuinely cannot get there, because every line of it is something you cannot change later without us. Three to five days.
4. The money and the delivery. Payment gateway connected on your account with you at the keyboard, shipping zones and rates, tax settings, and the notification emails rewritten — Shopify's defaults are functional and forgettable, and the order confirmation is the single email your customer actually reads. The statement descriptor gets set here too. Two to three days.
5. Test orders, properly. A successful card, a declined card, a card that triggers 3D Secure, a full refund and a partial refund, each checked all the way through — order created, stock decremented, email sent, money visible in the gateway. On a phone, not only on a desktop. One to two days.
6. Domain, launch and handover. DNS pointed, redirects in place if anything existed before, a walkthrough video and a short written guide covering the ten things you will actually need to do yourself. One to two days.

The tools, so this is not a black box — a Shopify Partners development store to build in before you pay for a plan, Shopify CLI for theme work, Judge.me or Loox for reviews, Klaviyo for email, PageSpeed Insights for the speed pass, and Figma when the design needs approving before it is built.

What moves the price

The range is $700–$4,000 over 2–4 weeks, and three things decide where a project sits in it.

How much catalogue there is. Twenty-five products is one job. Three hundred products with size and colour on each is a different job with the same name. Variants multiply — the count that matters is not products, it is product-variants.

How far the design goes past the theme. Configuring a good theme is fast and predictable. Rebuilding sections in code is neither, and it also makes the store harder for you to edit afterwards. We will always tell you which of your requests is a setting and which is a build.

Whether the content is yours or ours. If descriptions, image preparation and collection copy come to us finished, the build is a build. If we are writing and preparing them, that is a separate line with its own price, and it is usually the largest one.

Migration from another platform, subscriptions, or multiple currencies each push a project toward the top of the range or out of it into their own quote.

Where these builds go wrong

Products get typed in by hand. Twenty-five products with three variants each, entered one at a time through the admin, is three days of work and a guaranteed crop of typos in SKUs and prices that surface a month later as stock that does not reconcile. The CSV is not bureaucracy. It is the only version of the catalogue that can be reviewed before it is live and corrected in bulk afterwards.

The monthly bill is never mentioned until it arrives. A fixed build price is quoted, then during the project the store needs reviews, then email, then a better page builder. Each of those is an app and each app is a monthly charge. The owner sees that bill after launch and asks what it is, having believed they had paid for everything. This is avoided with three sentences in the quote naming the platform fee, an estimate of the app fees, and a commitment that no paid app gets installed without written approval.

The statement descriptor is left on the default. This is a single field in the payment settings and almost nobody sets it. A customer who bought from a candle brand sees an unrecognisable abbreviation on their bank statement six weeks later, does not recognise it, and calls their bank instead of you. That is a chargeback, and a chargeback costs the order, a fee, and the stock that already shipped. The same category of neglect covers order confirmation content and a support route in the footer — the customer who can reach you does not go to their bank.