On 26 August 2026, Shopify finished removing script tags and Additional Scripts from the Thank You and Order Status pages for merchants on Basic, Shopify and Advanced. Stores were upgraded automatically. Nothing on the storefront broke. Orders kept landing.
That is the problem.
Shopify Plus went through the same thing a year earlier, on 28 August 2025, when checkout.liquid and Additional Scripts were retired there. Shopify Scripts shut down permanently on 30 June 2026, after two extensions. Every deadline in this story has now passed, and the web is still full of countdown guides telling you how to prepare for them.
What actually happened, what to check this week, and what any of it has to do with the carts you are not recovering.
The timeline, so we are working from the same dates
- 28 August 2025: Shopify Plus lost checkout.liquid and Additional Scripts on the Thank You and Order Status pages.
- 30 June 2026: Shopify Scripts shut down permanently, after two prior extensions.
- 26 August 2026: non-Plus stores had script tags and Additional Scripts removed from the Thank You and Order Status pages by automatic upgrade. Removed, not migrated.
Source for all three: Shopify’s own developer documentation on the checkout.liquid deprecation, and the Shopify changelog covering the Scripts extension to 30 June 2026.
The failure is invisible by design
When the auto-upgrade ran, no page errored. No customer saw anything. Checkout worked. What stopped was every piece of third-party JavaScript that lived in Additional Scripts or in a script tag on the Thank You and Order Status pages.
In practice that usually means some combination of: a hand-pasted Meta pixel firing the purchase event, a Google Ads conversion tag, a Google Tag Manager container, affiliate and referral tracking, a post-purchase survey, a review-request trigger, a warranty or subscription script.
Nobody gets an alert when those stop. The symptom arrives weeks later, and it arrives in disguise. It looks like Meta is underperforming. It looks like our conversion rate dropped. It is neither. It is a measurement outage, and it degrades retargeting, attribution and the audiences that feed your cart recovery ads. A brand can run on bad data for a full quarter and spend real money reacting to it.
One important exception, because it changes what you should look for: server-side integrations were not affected. Klaviyo, for example, reads the Shopify API rather than the checkout page, so its Placed Order metric kept flowing. Same for pixels installed as app pixels through official sales channel apps. The casualties are specifically the things a developer pasted into a text box years ago and nobody has opened since.
Four checks to run today
In this order. Each one takes minutes.
- Purchase events against orders. In Meta Events Manager, set the range to the last 14 days and compare Purchase event volume to the order count in your Shopify admin for the same window. In the stores we run we treat a gap past 5% to 10% as worth investigating. If Purchase events dropped sharply in the last week of August 2026 and orders did not, you have found your problem.
- GA4 purchases against orders. Same window, same comparison, in GA4. If your GA4 purchase count is now a fraction of your Shopify order count, or zero, the tag was living on the Order Status page.
- Read your Customer events list. Shopify admin, Settings, Customer events. This is now the only place checkout tracking legitimately lives. If your Meta pixel is installed through the Facebook and Instagram sales channel it runs as an app pixel and was never at risk. If someone pasted the base code into Additional Scripts in 2021, it is gone and it is not coming back on its own.
- Place a test order and watch the network tab. This is the one that catches the quiet vendors: affiliate networks, post-purchase surveys, review triggers, upsell apps. Open developer tools on the Thank You page and confirm each vendor’s call still fires. Healthy is boring: every request you expect, and nothing 404ing.
If you run more than two or three stores, this is worth automating rather than remembering. A small monitoring agent that compares yesterday’s Shopify order count with yesterday’s Meta and GA4 purchase counts, and messages you when they diverge past a threshold, catches this class of failure permanently. Tracking outages are not rare events. This one was just unusually well signposted.
Custom pixels are the replacement, and they are not drop-in
A custom pixel is JavaScript you add through Shopify’s Customer Events API, which runs in an isolated sandbox rather than directly on the page. Shopify’s migration guidance gives two supported routes off Additional Scripts, an app pixel from an official app or a custom pixel, and the custom pixel is the one that covers code you used to paste in yourself. It is available on all plans, including Basic. That last part is a genuine improvement over the old world, where checkout access was a Plus privilege.
What it is not is a paste-and-go migration. Custom pixels run inside a lax-sandboxed iframe. Cookies, localStorage and sessionStorage are reachable only asynchronously, through the sandbox’s browser API, and there is no synchronous access to the page itself. Shopify’s own Help Center states plainly that not all third-party pixel functionality works inside the sandbox, which is a polite way of saying some of your vendors’ snippets will simply never work again in their current form.
There are six standard checkout events: checkout_started, checkout_contact_info_submitted, checkout_address_info_submitted, checkout_shipping_info_submitted, payment_info_submitted and checkout_completed. Note the second one. checkout_contact_info_submitted is the moment a shopper stops being anonymous traffic and becomes a contactable abandoner, which is the hinge the whole back half of this article turns on. For most consumer brands the six are enough to rebuild purchase tracking properly. For anything that relied on synchronous access to the page, it is not, and the honest answer is to move that logic server-side or drop it.
What Shopify’s conversion numbers actually say
Shopify’s checkout page carries a headline claim that agencies have been repeating for three years. Here is the footnote, verbatim, from shopify.com/checkout as accessed in September 2026:
“Shopify’s overall conversion rate outpaces the competition by up to 36% and by an average of 15%, based on a study completed in April 2023 in partnership with a Big Three global management consulting company.”
Four things in that sentence that never survive into the blog posts. The average is 15%, not 36%. The 36% was against the weakest comparator. The study is from April 2023, which makes it over three years old. And it was commissioned and paid for by Shopify, from a firm it declines to name, with a methodology it has never published.
We build on Shopify every week and we think its checkout is genuinely the strongest default in its class. Anyone quoting 36% as an independent finding either has not read the footnote or is hoping you have not.
Shop Pay
Shopify’s Enterprise blog, on 3 August 2026, published a set of Shop Pay figures from the same April 2023 Big Three study: up to 50% higher conversion than guest checkout, at least 10% higher than other accelerated checkouts, roughly 5% lower-funnel lift from merely offering it, and around 16% higher average order value ($119.37 against $102.65).
Directionally we believe it. We see the effect in stores we run. It is also not independently verified, and Shopify’s own pages do not agree with each other. shopify.com/checkout says Shop Pay is “3x faster”; the Enterprise blog says 2.5x. shopify.com/checkout says “100M+ Shop Pay users”; the Enterprise blog says “more than 250 million shoppers”. Those are two different definitions being used interchangeably in the same marketing quarter.
The omission that matters more than either: Shop Pay depends on Shopify Payments in practice, which as of September 2026 covers 40 countries and regions. India, Brazil, Nigeria, Indonesia and China are not among them. Merchants outside the supported list use a third-party gateway and pay Shopify an additional transaction fee, which as of September 2026 runs from 0.2% on Plus to 2% on Basic. The single biggest checkout conversion lever Shopify sells is geographically gated, and the marketing never says so.
One-page checkout took your funnel report
As of September 2026 we have not found a Shopify page that publishes an average conversion improvement for one-page checkout. Its enterprise blog gives named case studies instead, Stellar Eats at 3.5% and Hemlock and Oak at 7%, which is a different kind of claim. The averages attributed to Shopify did not turn up on any Shopify page we checked, and the larger lifts circulating in agency and app-vendor posts are single-store anecdotes without control groups. Treat them as advertising.
The under-reported cost is measurement again. Collapsing contact, shipping and payment into one page removed the step-level page views those steps used to generate. GA4’s default checkout funnel no longer maps to what your customers actually experience, and drop-off between the three stages stopped being free to measure. Analyzify and Littledata have both documented the mismatch. A lot of brands lost their checkout funnel report in 2024 and 2025 and have been reading a broken chart ever since.
The denominator nobody defines
Here is the part of cart recovery that almost no article states, and it is the part with the most money in it.
Every recovery-rate benchmark you have been quoted is a fraction whose bottom half is undefined.
An abandoned cart is any session where someone added a product and left. An abandoned checkout is a session where someone reached checkout and gave you contact details before leaving. Only the second one is contactable. Shopify’s native abandoned checkout email and Klaviyo’s Checkout Started flow both fire after the shopper hands over an email address. Klaviyo’s automatic Added to Cart event for Shopify does not need a separate onsite snippet: it syncs through a Shopify server pixel once behavioural event tracking is switched on in the integration settings. It records only for shoppers Shopify has identified, through a form, a login or details entered at checkout, so it does not widen the contactable pool on its own.
Meanwhile the number everyone opens with, Baymard Institute’s 70.22% average documented cart abandonment rate (a rolling average across 50 studies, last updated 22 September 2025), is measured across the entire journey. Your flow only ever sees a slice of it.
So the useful benchmark is the flow-level one. Klaviyo, across more than 143,000 abandoned cart flows:
| Metric | Average | Top 10% |
|---|---|---|
| Open rate | 50.5% | 65.34% |
| Click rate | 6.25% | 13.33% |
| Placed order rate | 3.33% | 7.69% |
| Revenue per recipient | $3.65 | $28.89 |
Klaviyo’s dataset is from 2023 and, as of September 2026, remains the latest public flow-specific cut we can find. A separate and broader Klaviyo cut, covering abandoned cart emails rather than whole flows, puts the click rate at 6% average and 11.3% for the top decile. Different report, different denominator, not a contradiction. Note what Klaviyo does not publish: what share of abandoners are reachable at all. Nobody publishes that, because it is different for every store.
Work out your own reachable pool
You can compute it from your own admin. Take a clean 30-day window and pull four numbers.
- A: sessions that added to cart (Shopify Analytics, conversion over time)
- B: orders placed
- C: abandoned checkouts recorded in your admin, which is to say abandoners who left an email
- D: how many of C are actually in your marketing-consented list
Your reachable share of abandonment is roughly D divided by (A minus B). We are not going to hand you a benchmark for that ratio, because we have not seen one published that we would trust. Compute yours. Expect it to be smaller than you think, and that is the point: identification, not copywriting, is where the headroom sits. Optimising a third recovery email is optimising the visible slice of the problem, and it is a different job from the one most lifecycle email and SMS programmes are set up to do.
The EU ceiling is structural, not a skill gap
GDPR requires the marketing consent box at checkout to be unchecked by default. That means a European brand’s identified-abandoner pool is smaller by design than an American one’s, on identical execution. Shopify shipped a feature allowing merchants to pre-select the email consent checkbox by region specifically to navigate this, and explicitly notes that its guidance is not legal advice. Select no regions and you are on active opt-in everywhere.
Our position, and we are a Rome studio so we live inside this: treat abandoned cart messages as marketing, not transactional, and take the smaller pool. If your recovery rate trails a US competitor’s, that is one legitimate reason before anyone starts rewriting subject lines.
The top reason people abandon is a pricing decision
Baymard’s checkout abandonment reasons study, excluding the “just browsing” segment:
| Reason for abandoning at checkout | Share |
|---|---|
| Unexpected extra costs (shipping, tax, fees) | 40% |
| Delivery too slow | 20% |
| Did not trust the site with card details | 19% |
| Forced account creation | 18% |
| Checkout too long or complex | 17% |
| Site errors or crashes | 17% |
| Unsatisfactory returns policy | 13% |
| Could not see total cost upfront | 12% |
| Card declined | 10% |
| Not enough payment methods | 9% |
The widely circulated variant of this list, with 48% for extra costs, is an earlier cut of Baymard’s own research and not a third-party invention. The figure Baymard publishes for forced account creation is 24%. The figures above are the ones on Baymard’s current page, consulted in September 2026, so those are the ones to quote.
Read the top line honestly. Number one is not a checkout UX problem. It is a pricing and shipping decision made in a spreadsheet months earlier, and no Shopify feature fixes it. Shop Pay will not save a shipping table you are embarrassed to show at step one. Neither will the ad creative that carries your recovery ads. If 40% of your abandoners are leaving over costs they did not expect, the highest-value work this quarter is showing the true total earlier, not shortening the form.
What Shopify still does not let you do
Post-migration, the plan gating as of September 2026 is worth knowing precisely, because vendors blur it:
- Checkout UI extensions on Thank You and Order Status pages: available on Basic, Shopify and Advanced.
- Custom pixels: all plans.
- Code on the Information, Shipping or Payment steps: still not available outside Plus.
- The Checkout Branding API, which is what gives real visual control of the checkout itself: Shopify Plus only.
- Colour and font changes through the checkout editor: all plans.
- Native A/B testing: Shopify Rollouts, announced in the Shopify changelog on 5 June 2026, runs an experiment between two whole themes or two whole checkout and customer account configurations from the admin. Rollouts needs the Basic plan or higher and experiments need Grow or higher. It tests whole configurations, not individual checkout elements, and it does not cover headless or custom storefronts.
Agentic checkout: the 2026 story that did not happen
If someone is selling you an agentic checkout roadmap, check the date on their evidence.
In March 2026 OpenAI confirmed it was discontinuing Instant Checkout, launched in September 2025, because it was not generating conversions. CNBC reported that roughly 30 Shopify merchants were available via Instant Checkout. Walmart’s Daniel Danker told WIRED that checkout inside ChatGPT converted roughly three times worse than a click-through to walmart.com, even though ChatGPT brought new customers at about twice the rate of search. CNBC and Search Engine Land both cover the wind-down.
The protocols continue. As of September 2026 the Agentic Commerce Protocol from OpenAI and Stripe, and the Universal Commerce Protocol from Google and Shopify, are both still active. But the honest 2026 framing is this: AI is a discovery channel, not a checkout. Shopify’s own Q1 2026 data supports that reading, reporting AI-referred traffic converting around 50% higher than organic search with about 14% higher average order value. That is a discovery-side claim, and it is the one worth building for.
What we would actually do this week
- Run the four checks above. Fix any pixel that died on 26 August 2026 before touching anything else, because every decision you make until then is based on bad numbers.
- Rebuild purchase tracking as a custom pixel, and accept that one or two vendor snippets will not survive the sandbox.
- Compute your reachable abandoner pool. Write the number down. Revisit it quarterly.
- Move the total cost, including shipping, as early in the journey as you can stomach. That is the 40% problem.
- Check whether your checkout funnel report in GA4 still describes reality after one-page checkout. If it does not, stop reporting on it.
- If Shopify Payments is available where you are and Shop Pay is not enabled, enable it. It remains the strongest single lever available, on numbers that are Shopify’s own but consistent with what we see.
We cut four widely repeated figures from this article, on failed-payment recovery rates, card decline rates without an account updater, a Shopify merchant performance report that does not exist, and one-page checkout lift, because none of them trace past low-authority SEO blogs. If a figure does not come from Shopify, Baymard, Klaviyo or a named news outlet, it did not make the cut.
How we work on this
EDEUS runs Shopify stores, checkout instrumentation and cart recovery flows as an operating job, not an audit deliverable. Most of what is written above came out of doing the migration on stores we are responsible for. If you want a second pair of eyes, send us your last 30 days of Shopify orders against your Meta and GA4 purchase counts and we will tell you what your checkout stopped reporting in August 2026.
Frequently asked questions
Did the Shopify change on 26 August 2026 break my store?
No, not in any way your customers could see. On 26 August 2026 Shopify removed script tags and Additional Scripts from the Thank You and Order Status pages on Basic, Shopify and Advanced plans via automatic upgrade. Storefronts kept working and orders kept processing. What stopped was any third-party JavaScript that lived in those places, typically hand-pasted Meta pixels, Google Ads conversion tags, GTM containers, affiliate scripts and post-purchase surveys. The failure is a measurement failure and it does not generate an alert. Server-side integrations such as Klaviyo were not affected, because they read the Shopify API rather than the checkout page.
What replaced Additional Scripts on the Shopify checkout?
Custom pixels replaced Additional Scripts, added through Shopify’s Customer Events API. Shopify’s migration guidance gives two supported routes, an app pixel from an official app or a custom pixel, and the custom pixel is the one that covers code you used to paste in yourself. Custom pixels are available on all plans including Basic, which is an improvement on the old Plus-only access. They are not drop-in. Custom pixels run inside a lax-sandboxed iframe where cookies, localStorage and sessionStorage are reachable only asynchronously, through the sandbox’s browser API, and there is no synchronous access to the page itself. Shopify’s own Help Center warns that not all third-party pixel functionality works inside that sandbox. There are six standard checkout events: checkout_started, checkout_contact_info_submitted, checkout_address_info_submitted, checkout_shipping_info_submitted, payment_info_submitted and checkout_completed.
What is the difference between an abandoned cart and an abandoned checkout?
An abandoned cart is any session where a shopper added a product and left. An abandoned checkout is a session where they reached checkout and provided contact details before leaving. Only the second group is contactable. Shopify’s native abandoned checkout email and Klaviyo’s Checkout Started flow both fire only after an email address is captured, and Klaviyo’s automatic Added to Cart event for Shopify, which syncs through a Shopify server pixel and needs no onsite snippet, records only for shoppers already identified by a form, a login or details entered at checkout. This is why recovery-rate benchmarks are misleading unless the denominator is stated.
Does Shopify checkout really convert 36% better than other platforms?
No, not as an independently verified finding. That figure is Shopify’s own commissioned number and it is usually quoted incompletely. The footnote on shopify.com/checkout, as accessed in September 2026, reads: “Shopify’s overall conversion rate outpaces the competition by up to 36% and by an average of 15%, based on a study completed in April 2023 in partnership with a Big Three global management consulting company.” So the average is 15%, the 36% was against the weakest comparator, the study is from April 2023, and the methodology has never been published. It is not independent research.
Does Shop Pay actually increase conversion rate?
On Shopify’s own numbers yes, and the caveat is that they are Shopify’s own numbers. Its Enterprise blog, on 3 August 2026, reports Shop Pay converting up to 50% better than guest checkout, at least 10% better than other accelerated checkouts, around 5% lower-funnel lift from simply offering it, and roughly 16% higher average order value at $119.37 against $102.65, all from the same April 2023 study with an unnamed Big Three consulting firm. They are not independently verified, and Shopify’s own pages disagree on Shop Pay’s speed (2.5x versus 3x) and network size (100M+ users versus 250M+ shoppers). Shop Pay also depends on Shopify Payments in practice, which as of September 2026 covers 40 countries and regions, excluding markets such as India, Brazil, Nigeria, Indonesia and China. Shopify documents activating Shop Pay with a third-party gateway only for stores located in Australia, France and the United States.
What is a realistic abandoned cart recovery rate in 2026?
At flow level, Klaviyo’s benchmark across more than 143,000 abandoned cart flows is a 50.5% open rate, 6.25% click rate, 3.33% placed order rate and $3.65 revenue per recipient, with the top 10% reaching 65.34%, 13.33%, 7.69% and $28.89. That dataset is from 2023 and, as of September 2026, is still the latest public flow-specific cut. Any recovery rate quoted against total abandoned sessions is not comparable, because Baymard Institute’s 70.22% average abandonment rate (50 studies, updated 22 September 2025) counts everyone, while your flow only reaches abandoners who left an email and consented to marketing.
Should I build for agentic or AI checkout in 2026?
Not as a checkout. In March 2026 OpenAI confirmed it was discontinuing Instant Checkout, launched in September 2025, because it was not generating conversions, with CNBC reporting roughly 30 Shopify merchants available via the feature and Walmart’s Daniel Danker telling WIRED that checkout inside ChatGPT converted roughly three times worse than a click-through to its own site. As of September 2026 the Agentic Commerce Protocol and the Universal Commerce Protocol both continue. The practical 2026 position is that AI is a discovery channel: Shopify’s Q1 2026 data reports AI-referred traffic converting around 50% higher than organic search with about 14% higher average order value.
Shopify and the Shopify logo are trademarks of Shopify Inc., used under their trademark guidelines. EDEUS Studio is a Shopify partner. Photography by EDEUS Studio.
Content created with the support of artificial intelligence
