Changelog

Each product now has a place to record how outside systems identify it — Amazon’s ASIN, Whole Foods’ UPC, Target’s TCIN, and so on — kept in one registry per product rather than scattered across integration configs. On the Relationships tab of a product, a new grid lets you add, edit, and remove these identifiers, with a Linked / Unlinked badge so you can see at a glance whether an identifier is already attached to a specific connected integration or is sitting in the registry waiting to be linked. Adding one pulls the partner name straight from the connected integration you pick, previewing exactly how it will be stored once linked, and duplicate entries surface a clear message instead of a silent failure.

From the integration side, a new Link External IDs action on the integration’s Actions menu opens a bulk-link modal that attaches all of an integration’s unlinked identifiers at once by partner name. Linking runs as a background job, so a big backfill returns to you immediately and completes on its own. Linked rows are click-throughs — opening the connected integration in a stacked side view without leaving the product page.

Underneath, this is a first-class registry with its own permissions and audit trail; every add, edit, remove, or link event is captured. Inbound order ingest already reads this registry to resolve products from partner-supplied identifiers — see the separate EDI 850 entry — so what you curate here directly drives how partner orders find their way to the right product.

Lead time is no longer required on a product cost. Saving a supplier cost or a purchase order cost tier without entering a lead time works — the value is stored empty rather than as a stand-in zero. If you do enter one, it still has to be a positive whole number.

A broad round of improvements across the EDI order path — from how orders come in, to how they’re matched against your catalog, to how outbound documents go out.

Inbound matching and units. Inbound EDI orders now match products through the same partner-identifier registry you maintain on the product page — UPC, ASIN, TCIN, whatever the partner sends — and convert cases or inner packs to eaches automatically, using a per-partner unit-of-measure mapping so quantity and price land correctly. Those mappings now live on the integration’s Data mapping tab, so you can add, edit, and remove them yourself instead of asking engineering to. A line that can’t be matched or converted doesn’t fail the whole order — it’s flagged and held so you can finish the mapping and let it through.

Inbound carrier mapping. Incoming orders resolve their requested carrier through the same custom and global carrier mappings other integrations use, so per-brand and global mappings you’ve set up for a partner’s carrier codes are applied on the way in. When no mapping matches, the raw external value is kept on the order so it can be resolved on the fulfillment.

Inbound B2B data. Orders created from an inbound document now carry their B2B data — label and routing flags — straight onto the order.

Outbound documents. Acknowledgments, ship notices, and invoices are flowing reliably again. Acknowledgments carry the right PO references, a class of documents that used to render empty on certain orders now renders correctly, and ship notices and invoices that had silently stopped going out after an internal migration are going out again — each ship notice tied to its specific shipment, and the invoice sent once, when the order finishes shipping. The per-partner configuration you already have still controls which documents are sent, and a document that fails to send now surfaces instead of failing silently.

Forward Shipment now works on EDI orders, and moved to the per-fulfillment action menu — click it and the same Advance Ship Notice dialog opens for review before sending. For every other integration the button behaves as before.

Behind the scenes, order shipments are backstopped so a near-simultaneous webhook can’t create a duplicate, the admin bulk-deliver task sends acknowledgments alongside ship notices and invoices, and the internal tool used to author and validate a trading partner’s template got a faithful live preview plus a dry run that shows exactly how an order would resolve, without touching a real order.

The unit of measure dropdown on the product page picked up a lot of new options. Weight now has Gram, Kilogram, Ounce, and Pound alongside the existing entries. Volume gains Fluid Ounce, Milliliter, Liter, and Gallon. Length and area cover Millimeter, Centimeter, Meter, Inch, Foot, Square Foot, and Square Meter. Container and grouping units — Bale, Bundle, Drum, Jar, Pair, Pallet, Roll, Sheet, Tin, Tray, and Tub — round out the list.

The Push Product Data action is reachable from the products list as a Bulk Actions → Push Product Data, alongside the existing entry on a product’s detail menu, and can now target Shopify as well as ShipHero. The dialog has an integration selector with a per-integration settings recap — Shopify shows the attribute mappings, ShipHero shows kit type and the rest — and pre-fills the toggle settings from each integration’s configured defaults. Settings categories with nothing visible are hidden, so the dialog stays focused on what matters for the chosen channel.

Product details can now be pushed from Endless into ShipHero. Run the new Push Products action and Endless looks each SKU up in ShipHero, then creates or updates the product with name, SKU, barcode, dimensions, country of manufacture, tariff code, customs description, value, and customs value. Bundles reconcile their kit components in the same pass. Inventory is never touched by the push — only the product record.

The ShipHero settings tile also got a small tidy: the kit-type choice (Virtual or Pre-Built) is now explicit, the wholesale and 3PL copy reads more clearly, and a few settings that were never wired to anything have been hidden.

The attributes list opens to the tag view now instead of the table view. The tag view is denser and more visual, and the table view is still one click away in the display options.

Two changes to how Shopify orders connect to products in your catalog.

The “Import products” action on a Shopify integration always works now. Before, the action was silently blocked on integrations whose product-import setting was off, so the only way to pull in products was to turn that setting on first. Whether to import is your call from the UI; nothing else gates it.

The same setting still exists, but it now controls a narrower behavior: whether Endless should automatically import a missing product when a Shopify order comes in referencing one. The default is off, so order ingest no longer creates products on its own unless you opt in. Integrations that already had this setting on stay opted in.

Regardless of that setting, an order item that arrived without a matching product is now linked to one as soon as the product shows up — created later, imported manually, or updated with the SKU the order is referencing — so an order that landed before its product did doesn’t stay orphaned. V2 orders behave the same way as V1.

Creating a new product — from the UI, a clone, an import, or an integration — used to do a round of work per warehouse location to seed its starting zero-stock inventory. On accounts with many warehouses and many locations per warehouse that added up to hundreds of database operations per product, which was the dominant cost of product import. It now does the same seeding in a single batch, so the time per product no longer grows with the number of warehouse-locations. Same set of locations gets seeded at zero stock; just much faster.

A round of performance work on the Products and Orders list pages and the order detail page. On large accounts these pages had been creeping up to tens of seconds, and on the largest order histories the Orders list was timing out entirely and falling through to an empty state. They’re all back down to milliseconds, regardless of how many products or orders an account has.

Order detail pages also return immediately again — the activity timeline fills in once the main page is up, instead of holding the whole page back until it’s ready.