Real-time visibility sounds like one thing. It's two.
The first is helping a buyer shop. They need to see what's worth ordering, build an order that may run to hundreds of lines, and move between catalog pages without losing their place. The second is protecting the order. When it's placed, it has to match what you can actually ship.
Most platforms try to do both with one mechanism: push a live number to every screen. That's the expensive way, and it still goes stale. There's a simpler split. Let buyers open the whole catalog and keep an always-on cart. Check available inventory when the order is placed.
If you run operations, IT, eCommerce or sales at an enterprise brand, this decides how much your team maintains, how much load your ERP carries, and whether a buyer can build an order without being interrupted. Here's what it means, what it saves, and where it costs you.
Job one: help the buyer shop. The buyer needs direction. Is this style available? Is it worth adding? A number that's close is enough. Nobody is committing to anything yet.
Job two: protect the commitment. When the order is placed, close isn't enough. The order has to be right at that moment, against the inventory in your ERP.
That's RepSpark's definition of real-time. Available inventory is checked when someone checks out to place an order. Every order is matched with your real inventory in your ERP, no matter how long the catalog or cart was open before the order was placed. It doesn't matter whether a retailer, a rep or your own sales team places it.
Job two gets a live check. Job one doesn't need one. Treating them as the same job is what makes real-time inventory so costly to build and run.
If you want the full definition and how it applies to retailers, reps and internal teams, read our post on what real-time inventory data actually means for an enterprise wholesale team. This post covers what that definition does for the catalog and the cart.
Start with the arithmetic. One style in 6 colors and 8 sizes is 48 SKUs. A line of 250 styles built that way is 12,000 SKUs. That's an illustration, not a customer figure. Now picture keeping a fresh available number behind each of those, for every buyer who might look at any of them. That's a lot of questions for your ERP to answer.
ERPs put limits on how many questions they'll take. Microsoft's Business Central documentation lists per-user limits of 5 concurrent OData requests and 6,000 requests per five-minute sliding window, with excess requests queued or returned an error. NetSuite caps concurrent integration requests by service tier and SuiteCloud Plus licenses, with base limits from 5 on Standard up to 20 on Enterprise or Ultimate. Oracle's companion documentation notes that allocating concurrency to one application reduces what's available to other integrations. These limits vary by ERP, tier and license, and they don't prove that any one design will hit them. But they show that your ERP is a shared resource, and an inventory feed competes with everything else for it.
Microsoft is direct about the e-commerce case. Its Dynamics 365 Commerce documentation warns that APIs which reach directly into Commerce headquarters to fetch on-hand quantities should not be used in an e-commerce environment "because of potential performance issues and the impact that such frequent requests can have on your Commerce headquarters servers". It points integrators to its inventory availability APIs, which work from channel-side estimates, for that job. Its own e-commerce inventory path uses a cache, 60 seconds by default, and recommends running product availability jobs hourly. Microsoft also says displayed availability numbers "aren't always 100-percent accurate in real time" and that many retailers prefer to show a status such as "Available" or "Out of stock" instead of a quantity.
Read that again. Even a live-looking number is a copy with a delay. You pay the overhead and still end up with a snapshot.
A caveat on the evidence. These are vendor documents about Microsoft and Oracle products, not studies of wholesale brands. We found no independent study that measures this overhead for wholesale. What exists is documentation, and it points the same way.
Wholesale orders aren't one-click purchases. They get built, reviewed, changed and finished, often by more than one person.
Research on B2B buying points the same direction. Forrester's State of Business Buying 2024 found that 13 people within an organization are involved in the buying decision on average. A Gartner survey of 632 B2B buyers, fielded August to September 2024, found buying groups ranging from five to 16 people across as many as four functions. McKinsey's 2024 B2B Pulse, drawing on nearly 4,000 decision makers, found customers use an average of ten interaction channels in their buying journey. None of these measure carts. They describe buying decisions in general. What they suggest is that orders pass through many hands and many touchpoints, so a cart that works only in one sitting won't match how people buy.
RepSpark Flow's always-on cart is built for that. The January 2026 release note says "Your cart now follows you across every page," and that buyers can log in and find their active draft "exactly as you left it." RepSpark's support documentation says changes save automatically and the order saves across all tabs in a session. A buyer can move from catalog to product page to a report without losing the order.
That design only works if building the cart doesn't wait on a live inventory answer for every line. Amazon's engineers made a similar choice for their shopping cart. Their 2007 Dynamo paper describes an "always writeable" data store, built so that "an add to cart operation is never lost," with conflicts reconciled afterward. The paper is about data storage, not stock checks, so it doesn't say when inventory should be verified. But it shows the same instinct: let the cart accept work now and settle the truth later.
We don't have a measured result for the always-on cart yet, so we won't claim one. It's a design for how buyers work.
Here's the model in three moves. Open thousands of SKUs for shopping. Let the buyer add what they want to the cart. Check available inventory against your ERP when the order is placed.
Flow's July 2026 "Unified Shopping Scope" update covers catalogs, assortments and bulk add. A buyer can work across a large catalog, and the check that protects the order waits for the moment of commitment.
Checking at that moment is a recognized design, not an exotic one. Microsoft's Dynamics 365 Commerce performs an availability check when an order is checked out, when the stock check is enabled. Salesforce's B2C Commerce documentation describes a model in which the platform "calls the external inventory system at checkout for a real-time stock level check". Adobe Commerce records a reservation when an order is placed.
To be fair, the same documentation shows other options. Salesforce also describes a real-time check at add-to-cart, and Microsoft's checkout check can be switched off. Those are mainstream choices, and they trade in more load or more rules about holding stock. So the real decision is where to spend your ERP's attention: on every click, or on the commitment. For a brand with thousands of SKUs, RepSpark spends it on the commitment.
Not a blank screen. By default, RepSpark shows the raw available quantity to all users. Brands can customize that, capping units for certain roles, such as B2B buyers, with a plus sign (for example, 200+) and showing red, yellow and green stoplight icons. RepSpark's July 2026 release notes also list an option to mask inventory.
Those numbers come from the inventory data your team syncs, so they're as current as the last sync. How often your team syncs is worth confirming with RepSpark for your setup.
The point is how to read the number. While a buyer browses, it's guidance. At checkout, it's the decision.
Checking at checkout moves a surprise to the end of the order. That's the cost of this model, and it's worth naming.
Baymard's usability testing with online shoppers found that 30% are "likely to simply abandon" when they're told a product is out of stock. It's research on online shoppers, not wholesale buyers, and it doesn't state a sample size for that figure, so treat it as a signal, not a benchmark. But the lesson carries over. A "no" that arrives late needs to be easy to act on.
Three things soften it. Show availability while buyers browse, so the checkout check rarely surprises them. Make the result of a failed check clear. In RepSpark's model, the order is checked against available inventory at checkout and the buyer is alerted if anything has changed. And give the buyer somewhere to put what isn't available yet. RepSpark's support site documents a way to start a separate draft order for items not available until a later date.
When you evaluate any platform, ask to watch the checkout moment itself, including what the buyer sees when availability has changed. A slide won't show you that.
NAOT moved independent retailers to self-serve ordering. Its case study describes retailers able to purchase 24/7 against live inventory, with the ERP connected so they could shop in-stock products, and a 15% revenue increase in the first three months. The case study ties that result to self-serve ordering as a whole, not to one feature.
Full Turn, an apparel brand, shows the ERP-first model. Its case study describes AIMS360 calculating available-to-sell inventory by ship date, with RepSpark surfacing it to reps and buyers through the API. The ERP owns the logic. RepSpark shows it.
Use these on your next evaluation, and on your current platform.
If a vendor answers with a sync interval, ask the question again. An interval tells you how fresh a copy is. It doesn't tell you when your order is matched.
IT. You stop feeding a live number for every SKU to every screen. Your ERP answers the question that matters, at the moment it matters, instead of answering all of them all the time.
Operations. Orders are matched with your ERP when they're placed. The cart can stay open, and the check still happens at checkout.
Sales. Reps and retailers can open the full catalog, build large orders and move around without being interrupted. Nobody has to call the warehouse mid-order to find out whether a style is available.
RepSpark connects to your ERP and treats it as the source of truth. RepSpark supports more than 20 ERPs and 35+ integrations, and the platform is built for enterprise brands with large catalogs, multiple brands and complex warehouses. Flow's always-on cart follows buyers across the catalog, product pages and reports. Available inventory is checked against your ERP when the order is placed.
To go deeper, read about RepSpark Flow, how ERP integrations work, and what enterprise brands get from the platform. IT teams can start with the IT overview, and operations teams with the operations overview. For recent releases, see Features and Updates. Our posts on real-time multi-warehouse order entry and a quick ERP data audit cover the ERP side in more detail.
If you want to see the checkout check against your own ERP, including what a buyer sees when availability has changed, schedule a demo.
In wholesale, it means available inventory is checked when someone checks out to place an order. Every order is matched with your real inventory in your ERP, no matter how long the cart or catalog was open. It describes when the check happens, not whether every number on every screen updates continuously.
Buyers see available quantity from the inventory data your team syncs to RepSpark. By default, the web and mobile apps show the raw available quantity, and brands can cap counts, add status icons or mask inventory. While browsing, the number is guidance. The check at checkout is what matches the order with your ERP.
An always-on cart follows the buyer through the shopping experience instead of living on one page. In RepSpark Flow, the cart follows the buyer across every page, changes save automatically, and the order saves across all tabs in a session. Buyers can move between catalog, product pages and reports without losing their order.
Because it's costly and the number still ages. ERPs set limits on concurrent and total integration requests, and Microsoft's documentation warns against APIs that reach directly into Commerce headquarters for on-hand quantities from an e-commerce environment because of performance impact. Microsoft's own e-commerce inventory path uses a cache and notes that displayed numbers aren't always accurate in real time. Checking at checkout puts the live check where the order is.
In RepSpark's model, available inventory is checked at checkout, when the order is placed, and the order is matched with your inventory in your ERP. The catalog can be open and the cart can stay on for as long as the buyer needs. The check happens at the moment of commitment.
It can, which is the trade-off of this model. Showing availability while buyers browse reduces surprises, and a clear result when a check fails helps buyers act on it. In RepSpark's model, the buyer is alerted if anything has changed when the order is checked at checkout. RepSpark's support site documents a way to start a separate draft order for items not available until a later date. Ask any vendor to show you the checkout moment, including what a buyer sees when availability has changed.
Ask whether browsing triggers live calls to your ERP, how the cart behaves across sessions, at which moments availability is checked, what the buyer sees when it has changed, whether an open cart reserves stock, and what you can control about what buyers see. If the answer is a sync interval, ask again. An interval tells you how fresh a copy is, not when your order is matched.
Sources: