Most online stores do not fail because of one big catastrophe. They fail because of six or seven small decisions made in the first month, each of which looked reasonable at the time.
This is the list of mistakes we run into most often, either when a shop that launched six months ago is not selling, or when we take over a store someone else built. For each one you get what it actually costs you, how to check whether your store has the problem, and what the fix looks like.
No fluff, no “optimise your funnel” advice. Twelve concrete things, in the order they usually hurt.
What is different going into 2026
Three shifts changed which mistakes are expensive and which ones you can survive.
Building a store got easy, running one did not. AI-assisted builders will give you a storefront in an afternoon. What they will not give you is a tax setup that matches your country, a shipping table that reflects real courier pricing, or a checkout that survives a Friday traffic spike. The gap between “a store exists” and “a store works” is wider than it used to be, and more people fall into it.
Analytics got blurrier. Between consent banners, tracking prevention in browsers, and ad blockers, a meaningful share of your traffic is now invisible to standard analytics. Decisions made on incomplete data look confident and are often wrong.
Buyers compare faster. The tab with your product is one of five open tabs. Anything that slows a decision down, a missing shipping cost, an unclear return policy, a checkout that asks for a phone number twice, sends them to one of the other four.
1. Choosing the platform before you know what you are selling
The platform question gets answered first, usually on the basis of a YouTube comparison, and then everything else is bent to fit it. That is backwards.
A store with 40 products, one warehouse and card payments has almost no technical constraints, and any mainstream platform will do. A store with variant pricing, B2B tiers, multiple currencies or an ERP to talk to has hard constraints, and the wrong choice there costs you a rebuild in year two.
What it costs: a migration. Moving a live store with order history, customer accounts and indexed URLs is the single most expensive thing on this list.
How to check: write down the three things your store must do that a generic shop does not. If you cannot name three, platform choice genuinely does not matter much and you should stop agonising over it. If you can, check each one against the platform before you commit.
The fix: decide on the basis of your constraints in two years, not your comfort this week. We compared the two most common options in Shopify vs WooCommerce, and looked at the question from the other direction in WordPress or Shopify.
2. Treating mobile as the second version
Most stores are designed on a wide monitor and checked on a phone at the end. That order shows in the result: filters that need two taps to reach, product images that require pinch and zoom, a sticky header that eats a third of a small screen.
The majority of store traffic arrives on a phone, and on many stores the majority of revenue does not, which is the actual symptom. A large gap between mobile traffic share and mobile revenue share is not a coincidence. It is a design report.
How to check: open your analytics, compare conversion rate on mobile against desktop. If mobile is less than half of desktop, the problem is the interface, not the audience.
The fix: design the narrow layout first and let the wide one be the adaptation. Test on a real mid-range phone on mobile data, not on a simulator over office wifi.
3. Product descriptions written for the catalogue, not the buyer
Manufacturer copy pasted into the description field does two bad things at once. It reads like a spec sheet, so it does not sell, and it is identical to the text on forty other stores, so it does not rank.
The fix: for each product answer the three questions a buyer actually has. Who is this for and who is it not for. What does it replace or improve. What is in the box. Write that in your own words. Even two original paragraphs per product beats a copied spec table.
Where to start: your top 20 products by revenue. The long tail can wait.
4. A checkout that asks for things you do not need
Every field in a checkout is a small request for effort, and some of them are requests for trust. Asking for a phone number without saying why, requiring account creation before purchase, or revealing shipping cost only at the final step all push people out.
Cart abandonment is high on every store, that is normal. What is not normal is abandonment concentrated at one specific step, which is what you get when a checkout asks for something unexpected.
How to check: look at where people drop in the funnel. If one step loses far more than the others, that step is the problem and you can usually see why by walking through it yourself on a phone.
The fix: allow guest checkout. Show the full price including shipping as early as you can. Remove every field that is not needed to fulfil the order, and explain the ones that stay.
5. Launching with no SEO structure at all
SEO on a store is mostly structural, and structure is cheap before launch and expensive after. Category URLs, the relationship between categories and filters, how variants are handled, whether out of stock products stay or disappear: all of these are decided at build time and all of them are hard to change once Google has indexed them.
The fix: before launch, settle the URL structure, make sure every category page has real text on it, and decide what happens to a product page when the product is discontinued. Ahrefs has a solid primer on the mechanics in their ecommerce SEO guide.
What not to do: launch first and “do SEO later”. Later means redirects, and redirects mean lost rankings.
6. Ignoring speed until it is a plugin problem
Store speed is usually not one bad thing. It is an unoptimised hero image, plus four analytics scripts, plus a slider nobody uses, plus a theme that loads the full icon font for three icons, plus hosting that was chosen on price.
Speed matters twice: it affects rankings, and it affects whether someone waits around for your product page.
How to check: test a product page, not the homepage. The homepage is the page everyone optimises and the one fewest people enter through.
The fix: compress and correctly size images, remove scripts you are not reading data from, and pay for hosting that matches your traffic. Most stores get the biggest single win from images alone.
7. No trust signals above the fold
A visitor who does not know your brand is making a risk assessment in the first few seconds. Can I contact a human. Can I return this. Has anyone else bought here.
Stores often have all three answers somewhere, in a footer link, on an About page, in a policy nobody clicks. That is not the same as answering them where the decision is made.
The fix: put the return window, the shipping timeline and a real contact option on the product page itself. Use reviews with names attached. A physical address and a phone number that a person answers do more for conversion than any trust badge graphic.
8. Security treated as a launch task instead of a routine
The dangerous state is not an insecure launch, it is a secure launch followed by eighteen months of nobody updating anything. Outdated plugins are the most common way small stores get compromised, and the damage is rarely dramatic enough to notice quickly. Injected spam links, a skimmer on the checkout, a redirect that only fires for mobile visitors.
The fix: automatic backups stored somewhere other than the server itself, a monthly update routine with someone’s name on it, and two-factor authentication on admin accounts. Test a restore once. A backup you have never restored is a hope, not a backup.
9. No plan for the questions that arrive after the sale
Pre-sale questions get attention because they obviously affect revenue. Post-sale questions get ignored, and they are the ones that produce refund requests, chargebacks and one star reviews.
The fix: write the five emails you will send over and over before you need them. Order confirmed, order shipped, delivery delayed, how to return something, how to reach a human. Having them ready is the difference between a two minute reply and a three day silence.
10. Launching without walking through a real purchase
Almost everyone tests that the checkout completes. Far fewer test what happens afterwards. Does the confirmation email arrive, and does it land in inbox or spam. Does the invoice show the right tax. Does the courier integration produce a label with the right weight. Does a refund actually return the money.
The fix: before launch, buy something from your own store with a real card, at full price, on a phone, from outside your office network. Then refund it. You will find at least one thing.
11. Building on rented ground without knowing it
Plenty of store owners discover during a dispute that the domain is registered to their former developer, that the hosting account is in an agency’s name, or that the theme licence is not transferable. None of that is visible while things are going well.
The fix: check today whose name is on the domain, the hosting and the payment gateway account. We went through the three things people most often assume are theirs in who actually owns your website.
12. Budgeting the build and forgetting the year
The build quote is the number everyone compares. It is also the smaller number. Hosting, domain, plugin and app subscriptions, payment processing fees, SSL, backups, updates and the inevitable small changes add up to a recurring cost that frequently exceeds the original build within two years.
The fix: before you commit, write down the twelve month total, not the launch figure. We broke the recurring side down in what a website really costs per year, and covered build pricing in ecommerce website cost.
A pre-launch checklist you can actually run
Go through this in one sitting. It takes about an hour and catches most of the twelve.
- Buy a product from your own store, on a phone, on mobile data, with a real card.
- Refund that order and confirm the money moves.
- Check the confirmation email in Gmail and in Outlook, and check the spam folder.
- Open a product page, not the homepage, in a speed test.
- Confirm shipping cost is visible before the final checkout step.
- Confirm the return policy is readable from a product page in one click.
- Search your own brand name and check what the result looks like.
- Confirm the domain, hosting and payment gateway are registered in your name.
- Confirm backups exist, are stored off the server, and restore correctly.
- Write down the full twelve month running cost.
Questions we get asked about this
Which of these twelve costs the most?
Platform choice and SEO structure, because both are decided once and are expensive to reverse. Everything else on the list can be fixed on a live store without much drama.
My store launched already. Which do I fix first?
Run the checklist above. Fix anything it catches that involves money moving incorrectly, then mobile conversion, then checkout fields. Leave structural SEO for a planned piece of work rather than a quick change, because half-finished URL changes are worse than none.
Is a slow store really losing sales, or is that just an SEO talking point?
Both things are true. It affects ranking, and it affects whether people wait. The reason to fix it is usually not dramatic, it is that image optimisation is one of the cheapest improvements available on most stores.
How much of this can I do myself?
The checklist, all of it. Descriptions, trust signals, checkout fields and the post-sale emails, all of it. Platform choice, SEO structure, speed work and security routines are where most people benefit from someone who has done it before.
Where to go from here
If you are still choosing a platform, start with Wix, Squarespace or WordPress. If your store already exists and has started to feel constrained, five signs you have outgrown your setup is the more useful read.
If you would rather have someone look at your store and tell you which of these twelve apply to it, get in touch. We will tell you what we would change and in what order, whether or not you work with us.