Home / Field guides / Websites & E-commerce
Field guide · Websites & E-commerceThe Field Guide to Building the Store Yourself
The sequence we work in, the decisions that are expensive to reverse, and the five patterns we are most often called in to repair. Written so that a careful owner could follow it without us.
Written to be used.
Some readers will finish this document and decide to build and run the store themselves. That is a good outcome, and we would much rather it happened before a fee than after one. The tools have genuinely improved: an attentive owner with a modern platform and a capable assistant can now get further, faster, than a small agency could a few years ago.
What follows is the order we work in and the reasoning underneath it, written plainly. Nothing has been withheld to make the work look harder than it is. Where a decision is genuinely reversible we say so, and where it is not we say that too, because knowing the difference is most of what experience actually buys.
You can generate the work. You cannot delegate the outcome.
What you can genuinely do yourself.
- A store that looks properly made
- The content layer, edited into voice
- Imagery prepared at catalogue scale
- Standard integrations, wired from documentation
- Your own numbers, read honestly
- Which decisions are reversible, which are not
- A catalogue where one decision changes reporting
- Arabic built alongside, never fitted over
- Rules that cannot be quietly reversed
If you sell a small catalogue from one location, in one language, to customers who pay by card, and you enjoy the building, run it yourself. Your first few hundred orders will teach you more than any firm can, and the fee is better spent on stock and photography. Come back when the catalogue, the languages or the fulfilment stop fitting inside one person's head.
Where the general answer stops being useful.
Not because it is wrong. Because it is general, and this is a contested problem against named competitors who were handed the same advice.
“Build me a Shopify store for a skincare brand selling in the UAE.”
All of it correct and none of it wrong. It is describing a store, rather than yours.
What the decision requiresThe first decision is not the theme, it is the product model: whether a 30ml and a 50ml are two variants of one product or two separate products. That single choice propagates into your filters, bundle pricing, shopping feeds, accumulated review counts, returns line-items and every URL you own, and reversing it after launch breaks the links and reviews you have already earned. Decide it against how you intend to discount and how you intend to report margin, not against how the product page happens to look.
“Write my product descriptions in English and Arabic.”
Correct as prose. The difficulty is that most of what needs translating is not prose.
What the decision requiresArabic commerce is an information-architecture problem before it is a language problem. Attribute values — colours, materials, sizes, fragrance families — must be translated once, as data, never per description, or the Arabic filters fragment: ivory rendered three ways becomes three colours, and a customer filtering by it is shown a third of your catalogue. Beyond that, Arabic buyers search in a plainer register than marketing copy is written in, and the words they actually type are usually shorter and more literal than the translation you would be proudest of.
“How do I improve my checkout conversion rate?”
Every item true and worth doing. It is also the list your competitors worked through last year.
What the decision requiresIn this region the loss usually sits below the interface. Split your funnel between payment attempted and payment authorised: a card declined at the issuer, or a 3-D Secure code that never arrives on a roaming or eSIM number, is invisible in most analytics and looks identical to a customer changing their mind. Authorisation rates also move with where your acquirer and merchant of record are domiciled, since a locally acquired transaction clears more dependably than a foreign-acquired one. That is a commercial arrangement rather than a design change, and it moves more revenue than any button ever will.
What a year of doing it alone usually looks like.
- Redirects mapped section to sectionEarned addresses resolve to a homepageA complete old-to-new inventory, rebuilt after the fact
- A translation layer over finished EnglishMirrored layout, English filters, addresses sharedSeparating them properly means going back to the catalogue
- Every app small and justified separatelyProduct page loaded with third-party scriptsFast on the founder's laptop, slow on customers' phones
- Store, point of sale and spreadsheet disagreeAn oversell, cancelled by handRefund is the small part; the customer never returns
- Sale price entered over the originalFeeds report a price movement that never happenedMargin history cannot be reconstructed; evidence gone
If you run it yourself, run it in this order.
- Before designWrite the order lifecycle downMost trouble is a state nobody decided, found too late
- Early onModel the catalogue before the themeEvery filter, feed and report depends on that vocabulary
- Before the buildSettle commercial rules in writingThese are checkout architecture; changing them later rebuilds the flow
- During the buildBuild Arabic in parallelFitted afterwards, it means going back to the catalogue
- Before launchMigrate, redirect, set a weight limitRedirects are proved against a full inventory of old addresses, never sampled
- After launchInstrument, then correctFirst corrections are to data and commercial rules, not design
No one can guarantee AI answers — we improve the signals that earn them.
Six workstreams, drawn as one machine.
A store works when six things are true at the same time.
What the PDF adds to this.
- —The six workstreams in full — what each one actually contains, not just what it is called.
- —One line item opened into twelve decisions — the part that decides the result.
- —The shape of a mandate — scope, cadence and what the work honestly asks of you.
- —The written argument behind every diagram on this page, in full.
Run it yourself.
Or hand it to someone whose name is on the result.
The tasks became cheap; who answers for the number at the end of the quarter did not change. We will tell you honestly if you are not a fit.
Request a Growth Audit