In short
In a small store every order counts in your numbers, so a few missing or doubled orders change the picture more than they would for a big store (part 1 of this series shows how few visitors a small store's tests have to work with). Check your tracking before the sale, while there's still time to fix it.
Who this is for: owners of small and new Shopify stores who want to trust their numbers over Black Friday and Cyber Monday, whether or not they run a test.
The short answer
- Check the three parts of tracking: the script on your pages, the pixel in checkout, and the order record Shopify sends after payment.
- Place a test order and follow it through. It's the only way to know the whole chain works.
- Know your consent setting. If shoppers who decline cookies aren't measured, your tracked orders will be lower than Shopify's. That's expected, as long as you know it.
- Leave your own visits out. In a small store, the owner's visits and test orders can be a noticeable share of the traffic.
- Fix things by Friday 13 November. After the 16 November freeze, don't touch tracking.
The 2026 timeline
| Date | What to do |
|---|---|
| By Fri 23 October | Run this checklist once. Fix anything broken. |
| By Sun 25 October | Start any test you plan to run, so it has three weeks before the freeze. |
| By Fri 13 November | Run the checklist again, after your sale banners and discounts are set up. |
| Mon 16 November | Freeze: no changes to apps, the theme or tracking settings. |
| 26 to 30 November | Check once a day that orders are still being recorded. |
Why tracking breaks before a sale
Most tracking problems in small stores aren't mysterious. They come from changes made for the sale:
- A new theme or a duplicated theme published without the app embeds switched on.
- A new app, popup or cookie banner that changes when scripts are allowed to run.
- A new checkout setting or payment method.
- The owner testing discounts with real-looking orders that end up in the numbers.
None of these break the store visibly. Orders keep coming in. The numbers just quietly stop matching.
The checklist
1. The app embed is on in your live theme
Testing and analytics apps on Shopify usually run through an app embed: a switch in the theme editor (Online Store, Themes, Customize, App embeds). App embeds belong to a theme. If you publish a different theme, or a copy you made before installing an app, check its app embeds too.
- Open the theme editor for your live theme, not a draft copy.
- Check that each tracking or testing app you rely on is switched on.
- If you publish a new sale theme, repeat this check the same day.
2. The web pixel reports checkout
Your theme can't see the checkout pages. Shopify's web pixels run in checkout and send events such as checkout started and purchase completed. Apps install their pixel when you install the app; you can see the list in Settings, Customer events.
- Check that the pixels of the apps you rely on are listed and connected.
- Don't add a second copy of the same tracking through custom code "just in case". Two copies count everything twice.
3. A test order goes all the way through
Place an order with a test payment method (Shopify's test mode, or the bogus gateway on a development store), or a real order you refund. Then check:
- The order appears in Shopify with the right total and discount.
- It appears in each app that should record it, once, not twice.
- It's marked as a test order where the app supports that, so it doesn't count in results. In Benchmyrk, Shopify test orders count only when Include test orders is turned on for a goal.
Do this on a phone as well as a computer. Mobile checkouts often use Shop Pay or another accelerated checkout, which is a different path.
4. Your consent setting is what you think it is
If you show a cookie banner, Shopify's Customer Privacy API tells apps whether each shopper allowed analytics. Apps that follow it don't measure shoppers who decline. That is correct behavior, but it means:
- Your apps will report fewer orders than Shopify. Orders from shoppers who declined still happen; they just aren't tied to a tracked visit.
- If your banner or its region settings change before the sale, the share of measured shoppers changes too, and a jump in "orders" may just be a jump in consent.
Write down your consent setup before the sale. Don't change it during the sale.
5. Your own visits are left out
In a store with 40 orders a week, the owner's own activity matters.
A store gets 2,000 visitors and 40 orders a week. In the week before Black Friday the owner places 4 test orders to check discount codes, and visits the store 50 times from a phone and a laptop. If none of it is excluded, test orders are 4 of 44 recorded orders (9%), and the owner's visits are about 2.4% of all visits. In a test, if those visits and orders land in one version, they can move its conversion rate on their own. Plain arithmetic; the store is illustrative.
Leave your own browsers, and anyone who works on the store, out of your measurements. Use test mode for test orders, so they're marked as tests.
6. Password pages and previews
New stores and development stores often have a storefront password. Tools that open your store to check pages (previews, screenshots, automated checks) can't get past it unless you give them the password. If your store is password protected until launch, decide on your launch date now: a store that opens on 20 November has no normal traffic to compare the sale with.
Mistakes to avoid
- Publishing a sale theme on 24 November. App embeds, pixels and banners all need checking after a theme change. Publish before the freeze, or not at all.
- Comparing Benchmyrk, Shopify and Google Analytics and expecting the same number. They count different things (visits vs sessions, paid vs placed orders, consented vs all shoppers). Know why they differ, then watch that the gap stays stable.
- Counting test orders as sales. Refund real test orders and use test mode where you can.
- Adding tracking code twice. A doubled purchase event makes every conversion rate look better than it is.
Checklist
- [ ] App embeds on in the live theme
- [ ] App pixels listed and connected in Settings, Customer events
- [ ] Test order placed on a computer and on a phone, recorded once in each app
- [ ] Test orders marked as test, or refunded
- [ ] Consent setup written down
- [ ] Your own browsers and devices excluded
- [ ] Storefront password stored where tools need it (if your store has one)
- [ ] Checklist run again by 13 November, after sale banners are live
Running this audit for client stores? See the pre-BFCM tracking and QA audit for every client store. For the general BFCM plan, see what to test before Black Friday, and for how Benchmyrk installs on Shopify, Shopify A/B testing.