Home › Resources › Checklists › Launch
eCommerce Launch Checklist
Most launch checklists are lists of features to switch on. The expensive mistakes are commercial: a shipping model that loses money on every distant order, a margin nobody calculated, analytics that were never verified. Forty-two checks, with the money questions first and the go-live QA last.
CHECKS
42 checks
TIME TO RUN
A day, spread over the final fortnight
LAST UPDATED
September 2026
LEVEL
Founder or project lead
The checks that matter most cost nothing to run and are almost never on a launch plan.
- Before you start
Run this twice
Once about a fortnight out, while there is still time to change something, and once the day before you go live.
The first pass is the one that matters. Almost everything in the first two sections is expensive or impossible to change after launch — a shipping threshold set wrong becomes a promise customers have already seen, and a URL structure becomes a redirect map.
The second pass is pure QA: click everything, buy something, refund it. It takes an afternoon and it catches the failures that would otherwise be found by your first real customer.
- A landed cost figure per product, however rough
- Carrier rates for the zones you will actually ship to
- Access to the domain, DNS and whatever you build on
- One person who can say no to the launch date
SECTION 01
7 checks
The commercial model
The section most launch checklists skip. Every check here is cheap now and expensive to change once customers have seen a price.
Calculate landed cost per product, not invoice cost
Include duty, inbound freight and payment processing. Most first-time sellers overstate margin by several points here and price accordingly.
Model delivery cost by zone and weight before setting any shipping rule
Orders travelling far cost several times more to deliver. A flat national rate set on a light product quietly loses money on every heavy, distant order.
Set the free-delivery threshold from contribution, not from a competitor
The threshold has to leave contribution after carriage at the average order value it produces. Copying a larger competitor is how thin-margin stores fund their rivals.
Decide your returns policy and who pays for return postage
It is a pricing decision, not a customer service one. Free returns on a low-margin category can erase the margin entirely.
Confirm you know your break-even before you spend anything on ads
The break-even ROAS calculator takes two minutes. Launching paid media without it is guessing with money.
Check tax registration and collection for every region you will sell into
Thresholds vary and back-taxes are not forgiven because a launch was rushed.
Write the terms, privacy and returns pages before launch, not after
They are a legal requirement in most markets and a trust signal in all of them. A store with no returns page converts worse.
SECTION 02
7 checks
Catalogue and content
What is on the shelf, and whether anybody could find or understand it. Structural decisions here become redirects later.
Settle URL structure before a single product is published
Changing it after launch means a redirect map. Decide the pattern for products, collections and pages now and stick to it.
Create a collection or category page for every term people actually search
Not one per internal catalogue division. The categories people search for and the ones your supplier uses are rarely the same.
Write unique descriptions for your top twenty products by expected revenue
Supplier copy on every product gives search engines no reason to prefer you. Twenty is achievable before launch; two hundred is not.
Check every product has a price, stock level, image and shipping weight
A missing weight breaks rate calculation at checkout. Export the catalogue and sort by blank cells.
Add alt text to product imagery
Accessibility requirement, and the only way image search can read your catalogue.
Confirm images are compressed and sized for the slot they render into
Launch-day traffic is the worst time to discover the homepage ships four megabytes of hero image.
Write the About and Contact pages properly
On a brand nobody has heard of, these are the pages that decide whether a first order happens at all.
SECTION 03
7 checks
Checkout, payments and delivery promise
The part where money changes hands, and where an untested assumption costs you a real order rather than a hypothetical one.
Place a real order on a real card, then refund it. Test-mode passing is not the same as working.
Complete a real purchase with a real card, then refund it
Test mode does not exercise the same path. Do this on desktop and on a phone, as a guest and as a signed-in customer.
Show total cost including delivery before the final checkout step
Unexpected extra cost at the end is the most frequently given reason for abandoning a checkout. Showing it early costs you nothing and keeps the orders that were real.
Enable guest checkout
Forcing account creation before a first purchase removes orders from people who have never bought from you before, which at launch is everybody.
Quote a delivery window you can actually hit for every postcode band
An accurate three days beats an optimistic two. Promises you miss generate refunds and support load from your very first customers.
Test every payment method you have enabled, including wallets
An Apple Pay button that fails silently on one browser is invisible to you and fatal to conversion.
Confirm order confirmation and shipping notification emails send and read correctly
Check on a phone. Transactional email is opened more than anything else you will ever send.
Set up abandoned checkout recovery before launch, not after
It is the one flow that pays for itself immediately, and it needs live traffic to start learning.
SECTION 04
7 checks
Technical and measurement
Get this wrong and you will spend the first quarter unable to tell whether anything is working.
Install analytics and verify it reconciles to store revenue on your test order
Do it with the refund too. If the numbers disagree at launch they will disagree forever, and every decision after that rests on a wrong figure.
Verify the domain in Search Console and submit the sitemap
Do it on day one so you have data from the beginning rather than from whenever somebody remembers.
Confirm the staging or development site is not indexable
A crawlable staging copy competing with your live store is a common and entirely avoidable launch failure.
Remove the site-wide noindex you set during the build
Every platform has one. Forgetting to remove it is the single most common launch mistake there is.
Check HTTPS is enforced sitewide and no mixed content warnings appear
One insecure image on a checkout page is enough to show a browser warning at the worst possible moment.
Test on a mid-range Android phone over a throttled connection
Not the newest iPhone on office wifi. Measure Largest Contentful Paint on a product page and aim under 2.5 seconds.
Set up consent management appropriate to your markets
Retrofitting consent after launch usually means losing the first months of analytics data.
SECTION 05
7 checks
Go-live QA
The day-before pass. Somebody who did not build the site should do this, because the person who built it cannot see it any more.
Click every link in the header, footer and main navigation
Tedious, fast, and it catches the broken link that would otherwise sit there for a month.
View the site on a phone, a tablet and a desktop
Check the basket, the checkout and one product page on each. Most launch defects are layout failures at one breakpoint.
Proofread prices, and then proofread them again
A misplaced decimal is the one mistake you cannot undo once orders arrive.
Submit every form on the site and confirm somebody receives it
Contact forms that go nowhere are the most common silent failure on a new store.
Confirm 404s return a useful page with a route back into the catalogue
You will have broken links. What matters is whether they lose the visitor.
Check search on your own site returns sensible results for your top ten products
Internal search is used by high-intent visitors and is almost never tested before launch.
Agree who is watching on launch day and how they will be reached
A named person, a channel, and a decision in advance about what would justify rolling back.
SECTION 06
7 checks
The first fortnight
Launch is not the finish line. These are the checks that catch what only appears once real people are using the store.
Watch for failed payments and checkout errors daily for two weeks
Early payment failures are usually configuration, not customers. They are also invisible unless somebody looks.
Read every support message properly
The first fifty customer questions are the most valuable product research you will ever get, and they are free.
Check Search Console coverage for anything unexpectedly excluded
Launch-time indexation problems are cheap to fix in week one and expensive in month six.
Compare your actual delivery times against what you promised
If you are missing the quoted window, change the quote rather than hoping the carrier improves.
Confirm the abandoned checkout flow is sending and converting
It is the flow most likely to be misconfigured and the one that pays back fastest.
Record a baseline before you spend on acquisition
Conversion rate, average order value, abandonment. Without these you cannot tell whether any spend worked.
Resist adding apps and features for the first month
Every addition is script weight and a new failure point. Let the store be boring and stable while you learn what it needs.
- Getting it wrong
Six ways launches go wrong
None of these are technical failures. All of them are decisions made without the numbers.
Copying a competitor’s free-delivery threshold
Their margin, weight profile and volume are not yours. A threshold that works for a large competitor can be loss-making on every order you take.
Leaving the build-time noindex in place
The most common launch mistake there is. Weeks of wondering why nothing ranks, for a setting nobody thought to check.
Buying traffic on day one
Before analytics is verified and before break-even is known, ad spend is an expensive way to generate data you cannot read.
Promising a delivery window to win the sale
An optimistic promise converts slightly better and generates refunds, support load and bad reviews from your very first customers.
Launching with twenty apps installed
Each one is script weight and a failure point. Start minimal and add only what a real problem demands.
Treating the launch date as fixed
Somebody has to be able to say the store is not ready. If nobody can, the date will beat the quality every time.
If the commercial section is where you are least confident, stop and fix that before anything else on this list.
- Sequencing
What to do when
Ordered by how expensive each becomes if you leave it until after go-live.
When
What
Why then
Cost of delay
Before you build
Landed cost, delivery economics, shipping model, tax
These decide pricing, which decides everything downstream. Changing them after customers have seen prices is a repricing exercise.
Very high
During the build
URL structure, collection architecture, top-twenty product copy
Structural. Changing URLs later means a redirect map and a partial reset of everything you have earned.
High
Two weeks out
Analytics verification, checkout testing, consent, staging noindex
Needs a working site but must not be left to launch week, when there is no time to fix what it finds.
Medium
Day before
Full QA pass by somebody who did not build it
Fresh eyes on layout, links, forms and prices. An afternoon that saves your first customers from finding the bugs.
Low but embarrassing
First fortnight
Payment failures, support themes, indexation, delivery accuracy
Only visible with real traffic. Cheap to fix in week one, entrenched by month six.
Grows quickly
The pattern is simple: the commercial decisions are irreversible and the technical ones are not. Spend your scarce pre-launch attention accordingly.
- What next
Where to go from here
Four routes depending on where you are in the build.
01
STILL CHOOSING A PLATFORM
Platform choice decides what the next three years cost you. The comparisons set out where each one actually differs.
02
MOVING AN EXISTING STORE
Replatforming has its own failure modes, mostly around redirects and preserved rankings. Use the migration checklist instead.
03
LAUNCHED AND QUIET
Twelve checks with published thresholds, so you can see whether the problem is demand, conversion or measurement.
04
WANT A SECOND PAIR OF EYES
Five working days across six disciplines before you commit budget to acquisition.
- Questions
Launch FAQs
What founders ask in the last fortnight.
How long before launch should I start this?
Two weeks for the full list, and the commercial section much earlier than that — ideally before the site is built, because pricing and shipping decide structure. The day-before pass is pure QA and takes an afternoon.
What is the single most common launch mistake?
Leaving the build-time noindex in place. After that, a free-delivery threshold copied from a bigger competitor, which quietly loses money on every distant or heavy order.
Do I need all my products ready to launch?
No. Launch with the range you can describe properly and photograph well. A hundred products with supplier copy converts worse than thirty with real descriptions, and it takes longer to fix.
Should I run ads on launch day?
Not until analytics is verified against a real order and you know your break-even. Spending before both is buying data you cannot read, at the moment you can least afford it.
Is it worth delaying launch to get this right?
For the commercial section, almost always. For the technical polish, rarely — a live store with small defects teaches you more than a perfect store that has not opened.
What should I measure from day one?
Conversion rate, average order value, checkout abandonment and delivery accuracy against what you promised. Four numbers, recorded before you spend anything, so later changes have a baseline.
We are moving from another platform — can we use this?
Partly. The commercial and QA sections apply, but replatforming has its own failure modes around redirects and preserved rankings. The migration checklist covers those.
- Keep going
Related reading
CHECKLIST
Forty checks across product page, basket, checkout, trust and measurement — the next thing to run once you have traffic.
CALCULATOR
Work out what return on ad spend you need before you spend anything at all.
FREE AUDIT
Five working days across six disciplines, ranked into one constraint and a ninety-day sequence.
About to launch, or just did?
The free audit is most valuable in the first month, while decisions are still cheap to change and before acquisition budget has been committed.