Test your checkout before launch

The short version

Write down exactly what you're selling, have the real link opened from the real email on a device that isn't yours, try every option that changes the order, fill in personalisation the way real customers will, read the basket like a receipt, test payment and delivery safely, read every email, check the order record you'll make from, rehearse the day before, and sort what you find into blockers and things that can wait.

If you test your checkout before launch the way most people do, you'll probably miss the problem that matters.

Most of us check a launch in pieces. We look at the product page in the editor. We read the email draft. We glance at the social post. Each looks right on its own, so we tick "checked".

But a customer doesn't see pieces. They tap a link, look at a page, pick options, add it to the basket, pay and get an email. Then you make the thing from whatever your order screen says.

If any step in that chain changes, drops or garbles something, the whole order goes wrong.

Here's a made-up shop to follow through this guide. Number Nine Signs makes personalised house number signs in a small workshop in Tipperary. Customers choose a size, a background colour and a font, then type the number or house name they want. Each sign is made to order and dispatched in about ten working days, with fixing screws included.

They were launching a new range of house name signs to their email list, with a launch code for 10% off.

The bugs live between the pieces, which is exactly where piece-by-piece checks never look

Think about the handoffs in a single personalised order:

The email hands over to the product page
The product page hands over to the option picker and the text box
The options and text hand over to the basket
The basket hands over to delivery, discount and payment
Payment hands over to the confirmation email
The confirmation hands over to the order record you'll make the sign from

Each handoff is a chance for something to change without anyone noticing. A chosen colour reverts to the default. A long house name gets cut short. The discount applies to delivery instead of the sign.

Checking each piece separately can't catch this, because the problem isn't in any one piece. It's in the gap between two of them.

Maker reviewing a blank route card beside a plain parcel. — AI-generated editorial photograph. An illustrative launch-planning scene, not a live campaign.

Write down exactly what you're selling before clicking anything

Take two minutes to write one short paragraph:

"House name sign, large (40 x 15 cm), one per order, in navy, green or black, with a choice of two fonts and up to 20 characters. €79, or €71.10 with the launch code. Made to order, dispatched in about 10 working days. Fixing screws included. Delivery to Ireland and the UK."

This is your comparison point. Without it, a page can feel right simply because you've seen it so often. With it, you'll notice the moment anything doesn't match.

Make one route card for each link you'll share

If you're sharing more than one link, make a short card for each. Number Nine's email linked to the house name page with the code applied automatically. Their social post linked to the same page without the code.

Two routes, either of which could break. For each, note the exact link, the product it opens, the options, the price, any discount, the delivery rules and what the confirmation email should say.

Have someone else open the real link, from the real email

Your own devices know too much. They're logged into your shop, they remember your address, and they may show you draft pages customers can't see.

So send the actual launch email, as a test, to a friend. Ask them to open it on their own phone and tap the link, exactly as a subscriber would.

Then check what they see:

Is it the right page, not a draft, a preview or the homepage?
Do the name, price and timing match the email?
If the email promised the launch code would be applied, is it?

Number Nine's friend landed on the house number page, not the new house name page. The email had been drafted before the new page existed, using a placeholder link that nobody had updated.

That's the kind of bug that only appears when you follow the real link from the real email.

Maker holding a turned-away phone beside a plain parcel. — AI-generated editorial photograph. An illustrative link-checking scene, not a software screenshot.

Try every option that changes the money, the product or the timing

Go through each choice that changes something important: size, colour, font, quantity, made-to-order timing.

Pick each one and watch the page. Do the preview, price, and any timing note all change together?

Then focus on the least obvious combination. For Number Nine, that was the small size, in green, with the script font. The preview image didn't update to green when script was selected, because the script font previews had only been made for navy.

You don't have to test every possible combination. Focus on the options that change price, contents, suitability or timing, and write down what you tested and what you didn't.

Fill in the personalisation box the way real customers will, not the way you would

This is the step most launch checks skip, and for personalised products it's where the worst bugs hide.

You'll type a short, tidy test name like "Rose Cottage". Real customers type all sorts:

Long names: "The Old Schoolhouse Cottage"
Apostrophes: "O'Sullivan's"
Irish and other accented letters: "Tigh Shíle", "Bóthar na Trá"
All lower case or ALL CAPITALS
Trailing spaces, or an extra line break
Numbers and names together: "12 Ashgrove"
Nothing at all, by mistake

Try each one. Then follow it all the way through: the basket, the confirmation email and the order record.

Number Nine found two serious problems:

"O'Sullivan's" appeared in the order record as "O'Sullivan's". If they'd made the sign from that, the customer would have received a sign with code on it.
The fada in "Shíle" was dropped completely in the order export, leaving "Shile".

Both were fixed by changing a setting in how orders were exported. Neither would have been found with "Rose Cottage".

Also check what happens if the box is left empty. A clear message saying "Please enter the name for your sign" is far better than an order arriving with a blank sign.

Read the basket line like a receipt you're about to sign

Add the sign to the basket and compare the basket line with your one-paragraph description:

Then behave like a real customer: change the quantity, go back to the product page and return, remove the item and add it again. Check nothing changes or disappears.

On a phone, check that long personalised text doesn't get cut off in the basket in a way that makes the customer think it's been shortened.

Test the launch code the way people will actually use it

Try the code in the exact route subscribers will use. Then try it wrong:

Typed in lower case
With a space at the end
On a product it shouldn't apply to
After the end date, if you can set the date to test it

Check whether it applies before or after delivery. Number Nine's code was taking 10% off the delivery charge too, which wasn't the plan.

If you can't test the full rule without placing a real order, mark it "not checked", and consider launching without the code until you can.

Test payment and delivery without placing a real order

Use your shop's test mode or test payment method, if it has one. Don't place a real order you plan to refund, and never use a real customer's details or card.

During the test, enter an address you deliver to and check the delivery options, charge, time estimate and how tax is shown. Try an address you don't deliver to. Then leave a required field empty on purpose and read the error message.

At the final review step, check the order still shows everything you chose, including the personalised text.

A test shows what the shop does under the conditions you tried. If there's a step you can't test, write it down and check your platform's help pages, rather than assuming it works.

Read every email that follows as if you'd just spent your own money

Find the test confirmation email. It should clearly show:

The product, size, colour and font
The personalised text, exactly as typed
The price, discount and delivery charge
The delivery address
When it will be made and sent
How to get help, or how to correct a mistake in the text

That last one matters for personalised products. People make typos. Number Nine added a line: "Spotted a typo in your sign text? Reply within 24 hours and we'll fix it before we start making it."

Check the sender name, reply to the email to see where replies go, and tap every link.

Check the order record you'll actually make the product from

Customers see the confirmation email. You see the order screen, the packing slip or an export. That's the version you make the product from, and it's the one launch checks forget.

Open the test order in your shop's admin area. Print the packing slip. Look at any export you use.

Check that the personalised text, options and quantity appear clearly, in a form you can work from on a busy morning without guessing.

For Number Nine, the packing slip showed the colour and font as codes, "C2" and "F1", rather than names. On launch week, with thirty orders, that's how a navy sign ends up green. They changed the slip to show the names.

Maker comparing two plain product variants on a worktable. — AI-generated editorial photograph. An illustrative options check, not a product result.

Run the whole route again on a phone

Phones cause problems laptops don't:

The keyboard covering the personalisation box
An error message appearing below the visible screen
A preview image too small to check the text
A sticky button covering the delivery note
Tiny colour swatches that are hard to tap

Go through the route once more on a phone, typing a long house name with an apostrophe. Write down which phone you used, rather than claiming you checked every device.

Maker checking the contents of an open plain parcel and blank card. — AI-generated editorial photograph. An illustrative order check, not a real customer order.

Check stock and making capacity for a made-to-order launch

For a made-to-order product, "in stock" isn't really the limit. Your making time is.

Number Nine can make about fifteen house name signs a week. If the launch email brought sixty orders in a day, the "about 10 working days" promise would break for most of them.

So before launch, they set a limit: 40 launch orders, then a note on the page saying the next orders would take about three weeks. The shop's stock count was set to 40 to enforce it.

Check that your stock count, your making time and the delivery promise on the page all agree, and decide in advance what the page says when the limit is reached.

Maker working beside a blank turned-away laptop and plain parcel. — AI-generated editorial photograph. An illustrative safe-testing scene, not payment evidence.

Test the route for changing an order, not just placing one

Personalised orders get changed: a typo spotted, a colour reconsidered, a different font.

Test how that works before customers try it. Reply to the test confirmation email asking to change the text, and see where the reply lands. Check you can edit the order in your admin area, and that the change reaches the packing slip.

Number Nine found replies went to an old inbox nobody checked. That was fixed the day before launch, which is much better than the day after.

Make sure someone is watching the help inbox on launch day

Launch day brings questions at the same time as orders. Decide who answers them, and from which inbox.

Write three or four ready replies for the questions you expect: delivery times, changing the text, which font suits a short number, and what size suits a front door.

Maker considering a turned-away phone beside a plain parcel. — AI-generated editorial photograph. An illustrative follow-up scene, not a real email.

Rehearse launch morning the day before

Run the exact sequence you'll follow on launch day, the day before:

Make the new page visible, if you can safely hide it again
Send the launch email to two test addresses
Open the link from each, on two different phones
Walk the route to the confirmation email in test mode
Open the test order in your admin area and print the packing slip
Check stock and making capacity are set correctly

Time how long it takes. Note anything you had to look up. On launch day, you'll be following a route you've already walked.

Consider sending the launch email to a small group first

If you're unsure whether everything holds up under real orders, send the launch email to a small part of your list first, perhaps your most loyal customers, a few hours before everyone else.

Number Nine sent to 50 subscribers at 9am and watched the first orders arrive. Each one was checked in the admin area and on the packing slip. Only once those looked right did the email go to the rest of the list at noon.

It's not always possible, and it isn't a substitute for testing. But it turns any problem that slipped through into a small one rather than a launch-wide one.

Two makers checking a blank card, phone and plain parcel together. — AI-generated editorial photograph. An illustrative help-route scene, not a customer service record.

Sort everything you found into blockers and things that can wait

You'll have a list of problems. Sort them while you're calm.

Blockers stop the launch:

The email links to the wrong page
An option or the personalisation doesn't reach the order record correctly
The price or discount is wrong
Payment doesn't work
The confirmation email contradicts the page

Things that can wait:

A slightly awkward photo crop
A spare paragraph in the story
A social post you haven't finished

Keep a simple three-column record

What you saw	The risk	What you'll do
Email linked to the old house number page	Every subscriber lands on the wrong product	Update the link and resend the test
Apostrophe shown as code in order export	Sign made with the wrong text	Change export setting and retest
Launch code taking 10% off delivery	Lower margin than planned	Limit code to products only and retest
Colour and font shown as codes on packing slip	Wrong sign made on a busy day	Show names on slip

Mark anything you couldn't check as "unknown".

Two makers sorting plain parcels and blank cards into groups. — AI-generated editorial photograph. An illustrative prioritisation scene, not launch performance evidence.

If you can't fix a blocker, you have three honest choices

You can make the launch smaller, explain the limitation clearly, or delay.

A day's delay costs far less than thirty signs made with the wrong text. And when you fix something, start again from the original link, because one fix can break a later step.

Look at the first real orders an hour after launch

Once real orders arrive, open the first few in your admin area. Check the options, the personalised text, the discount and the delivery method look exactly right.

A test order shows what should happen. A real order shows what did. Number Nine's first real order had a house name in Irish with two fadas, and both came through correctly. That one look confirmed the fix had worked.

Maker checking a sealed plain parcel and blank route card at a tidy table. — AI-generated editorial photograph. An illustrative readiness review, not a launch result.

Your next step this week

Even if you're not launching anything yet, try this on one product that's already live.

Send yourself the last email you sent customers, and ask a friend to open the product link from it on their own phone.

Pick the trickiest option and, if it's personalised, type something with an apostrophe and an accent. Follow it to the basket, the confirmation email and your own order screen.

If the text, the options and the price come through exactly as entered, you've passed. If not, you've found the bug before a customer did.

Final checklist

Use these last checks before you publish or send the next version.

The exact public announcement link reaches the intended live product.
Product options update name, image, contents, price and timing together.
Cart lines preserve the selected variant and quantity.
A safe test or preview method covers delivery, totals and checkout errors.
Confirmation page and email match the order and fulfilment plan.
The route has been checked on a phone with a fresh session.
Blocking issues, unknowns and deferred improvements are recorded before sending.

Keep going

Marketing Campaigns / Practical guide

Small Product Launch Plan: Launch One Product Properly With One Page, One Email and an Order Limit You Can Actually Pack

A small launch should shrink the number of places you show up and the number of orders you take, not the facts a buyer needs before paying. Here's a plan built on one purpose, one fact card, one page, one tested route, one announcement and a packing limit you set before anyone clicks.

Maker holding a plain pencil over blank cards beside an unbranded product at a worktable.