Commerce & e-commerce
CoderConda reconciles stock between shop, marketplace and ERP with one data model and one log, in 21 days for €1,245 one-off. 6 postings, recorded in 3 systems: the normal case matches everywhere, 3 exceptions do not.
Stock per system in pieces, after each posting
| Posting | ShopShop | MarketplaceMarket | ERPERP |
|---|---|---|---|
| Normal case | |||
| Opening stock | 10 | 10 | 10 |
| Order in the shop, 2 pieces | 8 | 8 | 8 |
| Order on the marketplace, 1 piece | 7 | 7 | 7 |
| Exceptions | |||
| Return, 1 piece | 8 | 8 | 7 |
| Shop and marketplace restock at once. The ERP waits for the inspection. | |||
| Bundle, set of 2 pieces | 7 | 7 | 5 |
| Shop and marketplace count one set. The ERP books both parts. | |||
| Partial delivery, 1 of 2 shipped | 5 | 6 | 4 |
| The shop reserves both. Marketplace and ERP see only the piece that shipped. | |||
| Deviation from target | +2 | +3 | +1 |
| Target: 3 pieces available to sell | |||
Every system counts 1 to 3 more pieces as sellable than are free.
| Posting | ShopShop | MarketplaceMarket | ERPERP |
|---|---|---|---|
| Normal case | |||
| Opening stock | 10 | 10 | 10 |
| Order in the shop, 2 pieces | 8 | 8 | 8 |
| Order on the marketplace, 1 piece | 7 | 7 | 7 |
| Exceptions | |||
| Return, 1 piece | 7 | 7 | 7 |
| Uninspected does not count. The log records why. | |||
| Bundle, set of 2 pieces | 5 | 5 | 5 |
| The set is split into two parts and both are booked. | |||
| Partial delivery, 1 of 2 shipped | 3 | 3 | 3 |
| One shipped, one promised. Neither is free any more. | |||
| Deviation from target | 0 | 0 | 0 |
| Target: 3 pieces available to sell | |||
All three show 3. Every exception is in the log, with the rule it was booked by.
Example item, invented numbers. Which rule applies, I settle with you for each exception.
01 — Data ownership
I write down first which system has the last word on which field. When two disagree, this table decides.
| Field | Shop | Marketplace | ERP | Interface |
|---|---|---|---|---|
| Item number, dimensions | reads | reads | leads | no role |
| Copy and images | leads | reads | no role | no role |
| Price | reads | reads | leads | no role |
| Stock on the shelf | reads | reads | leads | reads |
| Commitments from orders | leads | leads | reads | reads |
| Available to sell | reads | reads | reads | computes |
Available to sell is the stock on the shelf minus the commitments from orders.
Example for a retailer with a shop, a marketplace and an ERP. Yours will look different.
02 — Proof
0
manual stock discrepancies
An integration synchronises shop, marketplaces and ERP through one shared data model.
Operational reconciliation is eliminated and data issues surface early.
REST APIs, ERP, Shopify, MariaDB, Automation
Case study: ERP-Integration03 — Price
| Item | Scope | Term | Amount |
|---|---|---|---|
| Tier 3, Sync | Orders and stock in sync between shop, marketplace and ERP, or up to five flows, each with alerts. | 21 days from access | €1,245 final price |
| Without an open interface | Export, intermediate layer and scheduled reconciliation instead of an API. | after the call | in the quote |
| Further systems | Additional marketplaces, a service interface, edge cases beyond the standard rules. | after the call | in the quote |
All prices are final prices. No VAT is charged under § 19 UStG. The term runs from the day I have access to your systems.
All tiers: Services04 — Questions
Usually not. A shop system only has to go when it cannot provide the data or the integration points it needs. That is settled before the proposal.
Then I build the route to it: export, intermediate layer, scheduled reconciliation. Slower than an API, but traceable and logged, and usually still faster than reconciling by hand.
The system, before your customer does. Every run writes a log, errors are named instead of swallowed, and an alert goes to you. A failed reconciliation can be repeated.
21 days from the day I have access to the systems (Tier 3). The effort sits in the exceptions: returns, bundles, partial deliveries. The number of fields hardly matters.
On request, as ongoing support. Without it you get a documented handover including access to the log, so your team has the same view.
05 — Appointment
Fifteen minutes on video. You describe the two systems and the gap between them, I tell you whether an integration is worth it and which one I would start with.
The calendar is hosted by cal.com.
Loading it opens a connection from your browser to cal.com (USA), which transmits your IP address; cal.com may set cookies when it loads. You can also go straight there instead.
