Industries
Your platform sells fine. It just does not know what is sitting in fourteen stores.
How it works today
At some point somebody set a fixed safety buffer of three per SKU to stop overselling. Nobody remembers who. Since then the last three pieces of every article have been unsellable online, and the e-commerce manager who lowers that number has customer service on the phone within a week about an order a store had already sold over the counter.
Fourteen shops full of stock the webshop cannot see. The connection sends one number per SKU with no location behind it, so ship-from-store and selling another branch's stock at the till have been on the roadmap for three years. The fixed buffer of three takes another three pieces off every article. CBS measured the gap in 2025: pure online retailers grew 7.4 percent, retailers with a store and a webshop 1.9 percent.
A nightly CSV export from the till system gets called realtime. The stores keep selling all day, so in the morning the webshop sells pieces that left the shelf yesterday. On bol an order lands on stock reserved for click and collect. Every cancellation counts towards the service norms, and the service norms decide your buy box. Customer service explains it, and the store manager takes the blame for stock he does not control.
In the ERP, in a spreadsheet, in the webshop and in the bol feed. Every channel wants its own attributes and image ratios, and a collection launch takes weeks because one mandatory attribute is always missing and the listing gets blocked. Since GPSR took effect on 13 December 2024 it is a safety question too: manufacturer details, the EU responsible person and warnings have to exist as structured fields and flow to every channel, in Dutch.
Returns are graded, credited and put back on stock by hand, usually only in the DC, so what a customer hands back at the till goes into a van first. The average return rate was 23.5 percent in 2025, and 40 to 50 percent on clothing. Handling one costs 10 to 20 euro and can swallow up to 35 percent of the gross margin on an order. The Returnless benchmark measures more than three weeks before it is sellable again.
What we build for this sector
Not a menu of products. These are the pieces that tend to pay for themselves first in this line of work.
Most retailers do not need this. You do when the configurator needs more than three options per product, still the Shopify ceiling even after the variant cap went to 2,048 in October 2025. Also when bills of materials and installation are part of the sale, when B2B and B2C share one storefront, or when the front end has to pass WCAG 2.1 AA. On top of your existing checkout.
Part of E-commerce & WebshopsMiddleware between the till system, the e-commerce platform, the ERP, the WMS or 3PL, the shipping platform and the marketplaces, doing the translation and the error handling. Almost everything here has an API: SW-Retail, Picqer, Monta, Shopify, Exact Online, the bol Retailer API. The design work sits in the budgets: Shopify's GraphQL Admin API gives 100 points a second as standard, so a full catalogue sync cannot be brute-forced through it.
Part of System Integrations & APIsA stock service that tracks physical, reserved and free stock per SKU per location, and decides per channel which number gets published using rules instead of a fixed buffer of three. It recalculates on every till receipt, every order and every goods receipt instead of overnight. Store stock becomes sellable, which is the precondition both of those roadmap items were always waiting on.
Part of Inventory & Order ManagementEnrich in one place, with per-channel field mapping, so a single change lands on every feed instead of in four separate files. Underneath it, price history per SKU per channel where the lowest price of the past 30 days is calculated rather than typed. Almost no standard platform stores that history, and the ACM has been supervising it since 19 September 2025.
Part of Data Platforms & PipelinesRoute on channel, warehouse or store, with the carrier attached. Return intake with grading at the counter: condition and reason recorded, credit issued automatically, and the piece written straight back onto that store's stock instead of after a trip to the DC. 65 percent of retailers now charge for returns, but 79 percent of shoppers walk away over a policy they dislike. Faster handling helps, higher fees do not.
Part of Business Process AutomationThe screen that does not exist yet: one place where the e-commerce manager sees and changes channel prices, stock rules, product data quality, return reasons and marketplace performance. bol had roughly 52,000 sellers in mid-2023 and 43,300 on 31 March 2026, so that channel now rewards whoever runs it rather than whoever is listed on it. Plus a store app with pick lists for click and collect and ship-from-store.
Part of Internal Tools & Admin PanelsSystems we connect to
The sharpest question in any connection here: do you publish physical stock or free stock, and which reservations count? A click and collect order, an open bol order and a return nobody has scanned yet are three different kinds of reserved, and most stock differences between till and webshop start there. One date is already fixed: iDEAL to Wero has to be migrated by 31 December 2027, merchants included.
This is not an exhaustive list, and we are not a reseller for any of it. If a system has an API, a database, or an export, we can usually work with it. If it does not, we will tell you that before you spend anything.
Rules you work under
Four of the five rules below ask for a field you do not have: the lowest price of the past thirty days, the manufacturer's address, the weight and material composition per article, and a reasoned decision per report. The regulators are Dutch even where the rule is European: the ACM, the NVWA and the ILT.
What changes
Today
Webshop stock comes from an overnight POS export, with the fixed buffer of three still in it. Click and collect orders arrive at the store by email and get printed and left at the till. Ship-from-store does not exist because store stock is invisible online. Roughly 120 bol orders a month are cancelled for stock that was already gone, which hits the service norms and the buy box. Returns are processed in the DC only.
After
A stock service becomes the owner of the truth. Free stock is physical minus reserved minus a rule-based buffer per channel and per article group, recalculated on every transaction rather than once a night. On top of that, a store app as a PWA on the iPad already in the shop: pick list for click and collect, pick and pack for ship-from-store with a Sendcloud label, and return intake with order lookup. Publication to the webshop and to bol runs on webhooks.
The result
The buffer of three disappears, so 14 store stocks come up for sale online, last three pieces included. Cancellations fall because the published number is correct. Stores get orders on a screen, and a return taken at the counter is back on stock the same day.
An illustrative scenario based on how work in this sector is typically organised, not a specific client.
Related work we have shipped
Questions from this sector
You should not, and we are not asking you to. The platform stays the storefront and the checkout, with the payment stack untouched. What we build sits behind it: stock logic across channels, product data, the return flow, B2B pricing. That is where the twenty-four apps came from in the first place, because every gap in that layer got filled with another subscription. If your problem genuinely is the storefront itself, a replatform is a different conversation and usually a more expensive one.
Treat it as a data project, not a design project. The C-Series is being phased out toward the E-Series, built on Ecwid, and historical data such as old invoices does not migrate across automatically. eCom also stays a separate system from Lightspeed Retail, so the stock and order link between webshop and till is yours to solve either way. Start by securing the history you must keep, then decide where the storefront goes. The same logic applies to Adobe Commerce, where 2.4.5 extended support ends on 11 August 2026.
Usually that turns out to be less absolute than it sounds. SW-Retail publishes a REST API, Picqer has REST plus webhooks, Monta has a REST API, and most Dutch POS vendors have some integration route. If there really is no API, we build around the exports and imports that do exist, and we tell you up front what that costs you: latency, no realtime reservations, and a reconciliation step. What we will not do is pretend a nightly file is a live connection.
It is yours. The obligation sits with the party offering the service, which is you, not with whoever wrote the theme. The European Accessibility Act has applied since 28 June 2025, it points at WCAG 2.1 AA, and you are only exempt below 10 employees and 2 million euro turnover. Repairing it now is cheap: most of it is forms, focus states, contrast and the size and colour pickers, and that is work you can plan. After the first complaint reaches the ACM it is a date somebody else sets.
Related sectors
Wholesale & Distribution
Ordering portals that read live customer prices and stock straight from Exact or Business Central, plus the 2BA and GS1 data behind them. For a sales desk that cannot hire.
Logistics & Transport
We build beside your TMS, not over it: koppelingen, cost per rit including the vrachtwagenheffing, a chauffeursapp that needs no training, and a portal that stops the status calls.
Manufacturing
Between your ERP and the shop floor: digital job tickets, machine data, nacalculatie per order. Hours that are in Ridder iQ, Isah or MKG the same day, and a margin per order the DGA sees for the first time.
Describe how the work flows now, including the parts held together by spreadsheets. We will tell you what is worth building and what is not.
Start the conversation