Shopify's Summer '26 Editions shipped over 150 updates on June 17, 2026, and the headline for brands is that AI merchandising went native. Product ranking, cart-aware upsells, and theme and checkout A/B testing now live in core Shopify, free, where you were paying apps for them.
- AI Collection Sort orders collections and search results by predicted conversion, not manual drag-and-drop or a paid sort app.
- Predictive cross-sell blocks surface cart-aware upsells with no separate upsell app installed.
- Native A/B testing for themes and checkout lands under Markets, so you can test without a CRO app subscription.
- The Universal Commerce Protocol is on by default, which makes your catalog readable to AI shopping agents out of the box.
On June 17, 2026, Shopify streamed its Summer '26 Editions and shipped more than 150 updates. Most of them are housekeeping. A handful are not. For a brand running on Shopify, the real story is that Shopify just moved merchandising logic you have been renting from paid apps into the core platform. AI Collection Sort, predictive cross-sell, and native A/B testing are now built in, and several line items on your app bill are about to look hard to justify.
I read these drops from both sides of the table. I built the Shopify Partner Program, so I know how apps get built and priced, and I have run merchandising and conversion across a nine-figure brand portfolio, so I know what actually moves revenue on a product page. When Shopify absorbs a feature category, brands tend to do one of two things: turn everything on at once and break their funnel, or ignore it and keep paying for software the platform now gives away. Neither is right.
This is the merchant-facing read. If you want the broader ranking of every feature in the drop, the Editions ranked by impact breakdown covers the full list. Here I am only going to tell you three things: what to turn on, what you can cancel, and how to measure the change before you trust it. The same playbook applied to the prior drop in the Spring '26 Editions read, and the discipline matters more this time because the native features touch conversion directly.
One hard deadline sits underneath all of it. Legacy Shopify Scripts retire fully on June 30, 2026. If any discount, shipping, or payment logic on your store still runs on Scripts, it stops working after that date, Editions or no Editions. I cover the migration in the Scripts sunset post, and it is the one item here that is not optional.
The Summer drop,
on one page.
Strip the 150 line items down to what changes how a brand sells, and you are left with a short list. Native merchandising, native testing, a generally available checkout toolkit on Plus, and an agent-readable catalog by default. Here is the slice that matters.
| Feature | What it does |
|---|---|
AI Collection Sort | Orders collection and search results by predicted conversion Replaces manual sort and most sort apps |
Predictive cross-sell | Cart-aware upsell blocks, no separate upsell app |
Merchandising insights | A native panel in admin showing what is and is not converting |
Native A/B testing | Split-test themes and checkout under Markets, no CRO app |
Checkout Components | Generally available for Shopify Plus Composable checkout surfaces, no checkout.liquid |
Universal Commerce Protocol | On by default, makes the catalog readable to AI agents |
Scripts sunset | Legacy Scripts fully retired June 30, 2026 |
The pattern is the one Shopify has run for years, now aimed at merchandising. It watches which third-party features the platform cannot live without, builds a good-enough native version, and ships it to every store. Brands win because the baseline goes up for free. Apps that sold the baseline lose. Your job as a merchant is to figure out which of your apps were selling the baseline and which were selling something the native version still cannot do.
AI merchandising
moved into the core.
Turn it on, carefully.
The centerpiece of the drop is native AI merchandising, and it is three features that used to be three separate app categories. AI Collection Sort orders your collection pages and search results by each product's predicted probability of converting, in real time, instead of you dragging products around or paying a sort app to do it on rules. Predictive cross-sell blocks read the cart and surface complementary items, which is the job an upsell app used to do. And a merchandising insights panel in admin tells you what is and is not converting, which is the reporting layer you were probably also paying for.
Turn it on, but turn it on the way you would test any change that touches the path to checkout, which is to say one surface at a time with a holdout. AI Collection Sort is the highest-leverage of the three because collection pages are where most discovery happens, so start there, on one or two collections, and watch conversion and revenue per session against the collections you leave on manual sort. If you want the numbers to judge it against, the 2026 conversion benchmarks give you the medians by category so you know whether a lift is real or noise.
"Native AI merchandising is not a button you flip and forget. It is a default you should beat or accept on evidence, the same as any other merchandising decision."
The honest caveat is that a one-size predicted-conversion sort is tuned for the average store, not for your brand's strategy. If you intentionally lead a collection with a hero product, a new launch, or a high-margin SKU rather than the highest-converting one, the AI sort will quietly undo that. That is not a reason to avoid it, it is the reason to measure it. Run it as a test, keep the surfaces where your judgment beats the model on manual sort, and let the model own the long tail of collections where you were never going to merchandise by hand anyway. If your product pages need work before any of this matters, the product page audit framework is the place to start.
Native A/B testing
removes the last excuse
not to test.
The quieter feature that may matter more over a year is native A/B testing. Under Markets, you can now schedule, gradually release, and split-test themes, checkout configurations, and customer-account settings without a third-party app. For years the reason most brands under a certain size did not run real tests was that a credible testing app cost real money and real setup. That excuse is gone.
This is the feature I would adopt fastest, because it changes how you treat every other change in this drop. Native AI merchandising, a new theme, a checkout tweak, a cross-sell block, all of it can now be tested against a holdout inside the platform instead of shipped on faith. The brands that win the next year on Shopify will be the ones that stop arguing about whether a change worked and start measuring it, and Shopify just removed the cost of measuring.
What you can finally
cancel, and what
you should keep paying for.
Here is the part that pays for itself. When Shopify ships a native version of a feature category, a slice of your app bill becomes hard to defend. But "native exists" does not automatically mean "cancel the app." The right question for each app is whether you were paying it for the baseline that Shopify now gives away, or for something the native version still cannot do.
| App category | Verdict | Why |
|---|---|---|
Basic sort apps | Cancel candidate | AI Collection Sort covers rule-based and conversion sorting natively. |
Simple cart upsell apps | Cancel candidate | Predictive cross-sell handles cart-aware upsell out of the box. |
Basic A/B test apps | Cancel candidate | Native testing under Markets covers themes and checkout. |
Advanced merchandising suites | Evaluate | Keep if you use bundling, segmentation, or rules the native version lacks. |
Search and discovery apps | Keep, for now | Deep search relevance and synonyms still beat the native default. |
Do not cancel anything on day one. Run the native feature against the app for two to four weeks with a holdout, then cancel only the ones where the native version holds its own. The apps worth keeping are the ones selling judgment and depth: bundle logic, segment-specific merchandising, real search relevance, advanced rules. The apps worth cutting are the ones that sold you the baseline. For the bigger picture of which apps earn their place by revenue tier, the tech stack by revenue breakdown is the reference I point brands to.
Your catalog is now
readable to AI agents.
Make sure it reads well.
The least visible change in this drop may be the most consequential over a few years. The Universal Commerce Protocol is now enabled by default on every Shopify store. UCP is the standard that lets AI shopping agents read your catalog and build carts, and Shopify co-developed it with Google with endorsement from a long list including Amazon, Meta, Microsoft, Stripe, Etsy, Target, and Wayfair. Translation: your products are now discoverable to AI agents out of the box, where before that took a custom build.
For a brand, this is upside you do not have to earn, but it is upside you can squander. When a shopping agent reads your catalog to answer a customer, it is reading your product titles, descriptions, attributes, and structured data, not your beautiful PDP design. A product with thin copy, missing attributes, and no structured data is invisible to the agent even though it is visible to a human browsing your site. The brands that win the agentic channel will be the ones whose catalog data is complete and machine-legible, which is a different muscle than CRO.
This is the same shift I walk through in the agentic commerce guide for brands, and the practical move is unglamorous: audit your top SKUs for complete titles, full attributes, accurate inventory and pricing, and proper product schema. UCP being on by default means the channel is open. Whether anything sells through it depends on whether your data is good enough to be picked.
Your 30-day plan
to adopt this drop
without breaking anything.
Here is the sequence I would run if this were a brand I advised. It is built to capture the upside, prove it before you trust it, and stop you from torching your funnel in the excitement of free features.
Week one. Confirm nothing on your store still runs on legacy Scripts, because that retires June 30 and is not negotiable. Turn on native A/B testing first, since it is the tool you will use to evaluate everything else. Do not change anything customer-facing yet.
Week two. Turn on AI Collection Sort on one or two collections, with a holdout, and watch conversion and revenue per session. Audit your top twenty SKUs for catalog completeness so the UCP agent channel actually works for you. Leave cross-sell alone for now.
Week three. Add predictive cross-sell on a limited set of products, measured against a holdout, watching AOV and contribution margin, not just conversion. Review the native merchandising insights panel for the collections you have not touched. By now you should have early signal on the AI sort.
Week four. Make the cancel decisions. Any sort, upsell, or basic testing app whose native equivalent held its own over the test window gets cut. Keep the apps selling depth the native version lacks. Document what you turned on so the next person who touches the store knows why. If conversion is still the bottleneck after all of this, the gap is usually elsewhere in the funnel, and store speed is where I would look next.
Shopify just raised the merchandising baseline for every store on the platform, for free. That is good news for brands and a reckoning for a category of apps. Turn on the natives, test them like you would test any change to your funnel, cancel what they replace, and keep what they cannot match. The brands that treat this as a measured upgrade will come out ahead. The ones that flip every switch at once will spend July figuring out what broke.
What brands ask me
about the Summer '26
Editions.
The features that change how you sell are native AI merchandising (AI Collection Sort, predictive cross-sell, and a merchandising insights panel), native A/B testing for themes and checkout, Checkout Components reaching general availability on Plus, and the Universal Commerce Protocol being on by default. The hard deadline underneath the drop is that legacy Shopify Scripts retire June 30, 2026. For the full ranking, see the Editions ranked by impact breakdown.
AI Collection Sort orders your collection pages and search results by each product's predicted probability of converting, in real time, replacing manual sort and most paid sort apps. Turn it on, but test it on one or two collections against a holdout first, because a one-size predicted-conversion sort can undo intentional merchandising like leading with a hero product or a high-margin SKU. Judge the lift against the 2026 conversion benchmarks.
Strong cancel candidates are basic sort apps, simple cart upsell apps, and basic A/B test apps, since native features now cover those baselines. Keep advanced merchandising suites with bundling or segmentation, and deep search apps with real relevance, because the native versions do not match them yet. Do not cancel on day one. Run the native feature against the app with a holdout for two to four weeks first, using the tech stack by revenue guide as a reference.
The Universal Commerce Protocol (UCP), now on by default, lets AI shopping agents read your catalog and build carts, so your products are discoverable to agents without a custom build. The catch is that agents read your product data, not your design, so SKUs with thin copy, missing attributes, or no structured data stay invisible. Audit your top products for complete, machine-legible data. The agentic commerce guide covers the playbook.
Yes, and it is the one item in this drop that is not optional. Legacy Shopify Scripts are fully retired on June 30, 2026, so any discount, shipping, or payment customization still running on Scripts stops working after that date. Migrate that logic to Shopify Functions before the deadline, ideally with a few days of buffer for fixes. The Scripts sunset post walks through what breaks and what to migrate.
Sizing up which Summer '26 features to adopt and which apps to cut?
I built the Shopify Partner Program and run merchandising across a nine-figure brand portfolio, so I read these drops from both the merchant and the app side. That is the view I bring to deciding what to turn on, what to test, and what to cancel.
Start a conversation See the case studies →A note on sources: the June 17, 2026 stream date, the 150-plus updates, AI Collection Sort, predictive cross-sell, the merchandising insights panel, native A/B testing under Markets, Checkout Components reaching general availability on Plus, the Universal Commerce Protocol being on by default (co-developed with Google), and the June 30, 2026 Scripts retirement are drawn from Shopify's Editions announcements and contemporaneous reporting on the Summer '26 drop. The read on what to adopt, cancel, and measure is mine, from running merchandising at scale and building the Partner Program.