seokopat · product
> product.open("stokmatic")
Stokmatic
Stokmatic is an e-commerce back office that keeps product data, stock, orders, finance and channel delivery in a single data model. Its centre of gravity is not the integration itself, it is the quality of the data that reaches the channel.
While a brand sells on one channel, product data sits in a spreadsheet and nobody complains. Open a second channel and the spreadsheet becomes two. By the third there is no spreadsheet at all, there are three versions of the truth that disagree: one stock figure in the panel, another in the warehouse, a third in accounting.
We kept running into this on the consulting side. The problem is rarely a missing integration, it is the quality of the product data. Pushing stock and pulling orders is the easy half. The hard half is translating the same product into each channel's own vocabulary, its own required attributes and its own image layout.
That is what Stokmatic was written for. A product is entered correctly once and the system produces the channel version of it. The first build grew out of the daily needs of a real export operation, so every feature worked in an actual business before it became a product.
The problem it solves
Three different numbers for the same stock
The panel says one thing, the warehouse another, accounting a third. Which one is right only becomes clear on counting day. In the meantime a sold out item keeps selling and the cancellation rate climbs.
Every channel wants the same product differently
One makes sleeve length a mandatory separate attribute, another refuses the product without an EAN, a third silently crops anything outside its image ratio. The same catalogue gets rebuilt by hand for each channel and a field is missed every time.
Revenue is visible, margin is not
When commission tiers, payment provider fees, shipping, returns and currency movement are not split per product, all you are left with is a top line. Which product actually earns and which one is being carried stays unknown.
Errors surface as customer complaints
A wrong price, a missing variant or a rejected feed row usually shows up two weeks later in a complaint rather than the same evening. By then the cost of fixing it has already been paid.
What sets it apart
One record, a version per channel
Product, variant, imagery and pricing live in one data model. The shape a channel wants is derived from that record. Adding a channel becomes a template definition rather than another pass over the catalogue.
Data quality is checked at the gate
If a required attribute is empty the system flags it, tries to derive it from product properties and marks the row when it cannot. It will not push a made up value through the feed, because a row that fails the channel's quality control comes straight back anyway.
Dry run first, live on purpose
Bulk synchronisation defaults to dry run. You see what would go out as a report before anything moves, and live mode is switched on deliberately. Accidentally pushing live data was designed out.
Alerts before the drift grows
Anomalies in stock, pricing, production and sales are reported once they cross a threshold. The goal is not another report, it is making the error that would surface in two weeks visible on the same day.
What is inside
The product is being prepared for launch. This list shows scope; order and detail can shift with the roadmap.
Product data
- variant structure at model, colour reference and SKU level
- EAN, size, colour code and material data per variant
- multilingual descriptions, product attributes and care information
- image library classified by the role of each shot
Warehouse and stock
- multiple warehouses, transfers between them, movement history
- counting and movement records by scanning barcodes with a phone camera
- automatic alerts for items falling below a critical level
- channel stock calculated with reservations and a safety margin
Orders
- orders from every channel in a single list
- line level status: pending, shipped, delivered, returned, cancelled
- channel source and transfer price stored on the order record
- returns and cancellations reconciled against stock
Finance
- multi currency, daily rate tracking and manual rate override
- commission tiers, payment fees, shipping and return costs
- profit and loss per product and per channel
- settlement reconciliation and invoice generation
Channel delivery
- template based integration architecture, one file per channel
- required attribute mapping and a missing field report
- EAN format and check digit validation
- dry run report listing every rejected row with its reason
How it works
- 01
The catalogue moves in
Existing product data is imported from a spreadsheet, CSV or directly from a database. Missing and inconsistent fields come out as a report on day one, so the cleanup is listed rather than guessed at.
- 02
The channel is mapped
The target channel's required attribute set is extracted and mapped onto your data model. Derivation rules are defined for the gaps, and whatever cannot be derived is left visible.
- 03
A dry run goes out
The feed is produced in dry run mode first and validated in the channel's test environment. Rejected rows are reported one by one with their reason and corrected.
- 04
Operations take over
Stock, order and price flows become automatic. The team runs the day from the panel and the system reports drift as an alert. Adding a channel is no longer a project, it is a template.
Who it fits
Good fit
- fashion and lifestyle brands with hundreds of SKUs and a colour and size matrix
- sellers active on several channels at once, such as Trendyol, Hepsiburada, N11, Amazon Türkiye, Çiçeksepeti, PttAVM and Pazarama
- exporters opening up to European marketplaces who need multilingual and multi currency operations
- manufacturer brands that own the chain from production through to sale
- teams that keep failing channel approval because of product data quality
Not a fit
- single channel shops working with a few dozen products
- businesses whose current panel is enough and whose channel count will not grow
- simple setups where pushing stock and pulling orders is all that is needed
- work that has to go live today, since the product is still being prepared for launch
Technical foundation
The application is built on Next.js and React in TypeScript strict mode. The data layer is Supabase on PostgreSQL, where every table is isolated by tenant and protected with row level security policies. It was designed API first, so anything possible in the panel is possible through the API. The AI layer handles translation, content enrichment and anomaly analysis; the decision stays with a person.
Where the product stands
Stokmatic is being prepared for launch. If something is blocked on the channel side today, write to us anyway: we can look at your current setup, help on the data quality side and let you know when the product opens. We are not naming a date, because we do not want to promise something we cannot keep.
Frequently asked
Is Stokmatic live?
Not yet, it is being prepared for launch. This page explains what the product is and what it aims at. Once it opens we will link from here to stokmatic.com.
Why not use an off the shelf integration panel?
Most of them push stock and pull orders, which is the easy half. The blockage happens where product data has to pass a channel's quality control, and that is the job of the data model rather than the panel. We wrote it because we could not buy what we needed.
Which channels are in scope?
On the Turkish side the target is the large channels such as Trendyol, Hepsiburada, N11, Amazon Türkiye, Çiçeksepeti, PttAVM and Pazarama. On the European side the route runs through the intermediary platforms that aggregate marketplaces. Because the architecture is template based, adding a channel does not change the core system.
Can we keep our current accounting or ERP system?
Yes. Stokmatic was designed API first and exchanges data over CSV, XML and API. In Turkey the e-invoice and e-dispatch flow is set up through a certified integrator.
What does the setup ask of us?
The step that takes the most effort is catalogue cleanup. Missing EANs, empty required attributes and images with an unclear role all surface in the first report. We can do this work or your team can, but it cannot be skipped, because channel quality control looks at exactly this.
Where does the AI sit and who decides?
It sits in translation, description enrichment and the anomaly patterns that are hard to express as rules. The model drafts, a person approves. Content that breaks brand rules, lands in the wrong language or falls below the confidence threshold does not reach the channel.
If you are blocked on the channel side
How many channels do you sell on, how many SKUs do you carry and where exactly does it jam. We can help with this before the product goes live.
Related work
Stokmatic makes the most sense inside these services.