If you only have half an hour before the Boxing Day rush, spend it on a Shopify app permissions audit. Go to Settings > Apps and sales channels, open each app, and read what it can actually do to your store. Shopify changed this page on 5 November to show app activity and flag apps it considers a security risk, so the information you need is finally in one place instead of buried in an install screen you clicked through eighteen months ago. Thirty minutes. Most Canadian stores find at least two apps with write access to customers or discounts that nobody on the team uses any more.
That matters more in late December than in March. Between Black Friday and the 26th, you cannot take the store down to investigate anything. A discount code that appears out of nowhere at 9pm on Christmas night is a problem you have to solve live, with your developer asleep and Shopify support queued.
How to check Shopify app permissions in about ten minutes
From your admin: Settings > Apps and sales channels. You get the list of installed apps, with any security notices Shopify has raised against them. Click into an individual app and you can see the data it has access to and the permissions it was granted at install.
Do it in this order and it goes quickly:
- Count the apps. Write the number down. On most stores we audit it is higher than the merchant guessed by three or four.
- For each one, answer two questions: what does it do for us today, and does it need write access to do that?
- Flag anything that touches customers, orders, discounts, or your theme. Those are the four that can cost you money or a privacy complaint.
- Flag anything you cannot name the purpose of in one sentence.
Also check Settings > Users for staff accounts and, separately, any custom apps built for you. Custom apps don't go through the app store review process, so whatever scopes your developer requested are the scopes they have, forever, including after the project ended.
Shopify app scopes explained, without the documentation voice
A scope is a named permission. read_customers lets an app see your customer records. write_customers lets it change or create them. The pattern holds across the API: read means look, write means act. Apps request scopes at install, you approve them in one click, and that approval persists until you uninstall.
The ones worth reading carefully:
- read_customers / write_customers — names, emails, phone numbers, addresses, order history. Under PIPEDA this is personal information you are accountable for, including after it leaves your store.
- write_discounts and write_price_rules — the app can create discount codes and automatic discounts. A misconfigured loyalty or upsell app can stack discounts in ways your margin model never assumed.
- write_orders — can edit, cancel, or create orders. Fulfilment apps and duty calculators legitimately need it. A review widget does not.
- write_themes — can inject code into your live theme. This is the one that leaves a mess behind.
- read_all_orders — access to your full order history, not just the last 60 days. It requires separate approval from Shopify and most apps genuinely don't need it.
Nothing here is sinister on its own. A subscription app with write access to customers and orders is doing exactly its job. The question is whether the app still has a job.
What a Shopify app security warning actually means
When Shopify flags an app on the apps index, it is telling you that something about that app's status or behaviour no longer meets the bar it sets for listed apps. In practice that usually means the app has been removed from the App Store, is no longer supported by its developer, or has been identified as handling data in a way Shopify is not comfortable with.
It is not an accusation that your store has been breached. It is closer to a notice that nobody is maintaining the lock on a door you left open. The practical response is the same either way: if the app is flagged and you are not actively using it, uninstall it today. If you are using it, find out whether the developer is still responding to support emails before you head into peak. An unmaintained app that breaks on 26 December is an app with no one to call.
Customer data, PIPEDA, and the apps you forgot share it
Every app with read_customers is a third party holding your customers' personal information. PIPEDA makes you responsible for that information when it is transferred to a third party for processing, which means you are expected to know who has it and under what terms. If you sell into Quebec, Law 25 adds obligations around transparency and, in some cases, explicit consent for how personal information is used.
The uncomfortable version: most merchants cannot name, from memory, every app holding their customer list. Quiz apps, giveaway apps, the SMS tool you trialled for one campaign, the abandoned-cart app you replaced but never uninstalled. Each of those pulled a copy of your customer data into someone else's infrastructure.
So while you have the apps index open, build a one-page list: app name, what customer data it reads, where the company is based, and whether you still use it. That list is what you reach for if a customer exercises their right of access, and what you need if you are putting a privacy policy in front of a Quebec audience. It takes twenty minutes once and then it is maintenance.
Who installed this app, and when
The app activity view on the apps index is the first place to look. Beyond that, Shopify keeps a staff activity record under Settings > Users, where app installs and removals appear against the staff account that performed them. If you are on Shopify Plus, the organization admin and the detailed activity logs give you a longer window and more of the surrounding context.
Why bother? Because the answer is often "a contractor in 2023 who no longer works with us." That is your cue to also check which collaborator accounts still have access, and whether anyone on the list left the company. We've opened stores where the agency that built the site three agencies ago still had a live collaborator account with full admin.
Removing unused Shopify apps safely, in the right order
Uninstalling is one click, which is the trap. Do these three things first:
- Export anything the app holds that you want to keep. Reviews are the classic loss. Uninstall a review app and the review content, which lives on their servers, is usually gone with it. Same for subscription contracts, loyalty point balances, and quiz responses.
- Check whether it created anything that outlives it. Automatic discounts, draft orders, metafields, script tags, shipping rate profiles. Discounts are the one that bites in December: an orphaned automatic discount can keep firing after the app is gone.
- Duplicate your live theme before you touch any app with theme access. Name the copy with the date. If the storefront breaks, you revert in ten seconds instead of debugging at midnight.
Then uninstall. Then, and this is the step most people skip, check whether the app actually cleaned up after itself.
Leftover theme code after you uninstall an app
Apps built on app blocks and app embeds disappear cleanly when you remove them. Older apps, or apps installed by a developer who hand-pasted a snippet into theme.liquid, do not. What's left is a script tag pointing at a domain that no longer serves anything, or a Liquid include referencing a snippet file that has been deleted.
Symptoms: a 404 in the browser console on every page load, a layout that shifts for no visible reason, a Liquid error on a product template that only appears for logged-in customers. Where to look:
- theme.liquid, in the head and just before the closing body tag
- product.liquid or main-product.liquid, for widgets that rendered on the PDP
- cart templates, for upsell and shipping-bar remnants
- Online Store > Themes > Edit code > Assets, for orphaned JS and CSS files
- Settings > Checkout, for scripts added to the order status page
Search the theme for the app's name and for its CDN domain. Remove the block, save, test the page, and test checkout. If you have more than one theme published across markets, do it on each, because theme edits do not propagate.
Which apps slow the store down, and the arithmetic
Apps hurt performance by adding JavaScript that the browser has to download, parse, and execute before the page becomes interactive. A well-built app adds 20–40KB and defers it. A badly built one loads a 300KB bundle synchronously in the head on every page, including the ones where it does nothing.
Here is the kind of sum worth doing. Say you have eleven apps and four of them are unused. Those four average CA$34 per month in subscription fees. That is CA$136 a month, CA$1,632 a year, for software nobody opens. If three of them also inject a render-blocking script of roughly 60KB each, you are shipping 180KB of dead JavaScript on every page view. On a mid-range Android over LTE in a Canadian suburb, that is measurable in your Largest Contentful Paint, and it is money you are paying to make the site worse.
Check the real weight before you blame an app: run a page through PageSpeed Insights, open the network tab, and sort third-party requests by transfer size. The culprit is usually one app, not the stack as a whole. If you want this monitored rather than audited once a quarter, SwiftStore scans the store's PageSpeed score, fixes what it can automatically, and tracks the number over time. For a deeper rebuild of what loads and when, our speed optimization work starts with the same third-party audit.
The Canadian-specific apps to look at twice this month
Three categories where the December stakes are higher here than elsewhere.
Shipping and rate apps. If you added a carrier fallback after the last Canada Post disruption, you may now have two rate apps both writing to your shipping profiles. Open both. Confirm which one is actually calculating rates at checkout and whether the other is silently overriding it. Test a real Boxing Day scenario: a Toronto postcode, a Nunavut postcode, and a US address.
Tax and duty apps. GST/HST and PST differences across provinces are usually handled natively, but third-party tax apps and cross-border duty calculators request order write access and can modify line items. If you ship to the US, confirm your duty and de minimis logic is still doing what you think after any 2025 rule changes you absorbed earlier in the year. A customer who gets an unexpected carrier invoice in January files a chargeback, not a support ticket.
Discount, loyalty, and gift card apps. Boxing Day is a discount event layered on top of gift card redemption, and gift cards are at their annual peak on 26 December. If a loyalty app and a promotion app both have write_discounts, work out what happens when a customer applies a Boxing Day code to a gift-card-funded order with points attached. Test it. Don't model it.
Freeze the stack on 1 December
Finish the audit, remove what has to go, then stop installing things. No new apps between 1 December and the first week of January, no exceptions for the one that promises a conversion lift. Every install in that window is an untested change to a store that is taking orders.
If something genuinely has to go in, install it on a duplicated theme, test it there, and publish the theme rather than installing straight to live. Takes an extra fifteen minutes and has saved more than one Boxing Day.
Two things to do this week: open the apps index and deal with anything Shopify has flagged, and export your review content before you uninstall anything that holds it. If you want a second pair of eyes on the stack, scopes, leftover theme code and all, our team runs a free store audit, and we work with Canadian merchants on exactly this kind of pre-peak cleanup.


