4.9
Custom services written against the commercetools APIs, plus the connector work around them, mapped field by field before anyone puts a number on the build.
Official commercetools partner · 600+ in-house specialists

Four patterns carry every flow: a ready connector, an API extension, a subscription, or a custom microservice. Picking the wrong one is what makes these estimates wrong, and the choice is made per flow rather than per system.
Services written against the commercetools APIs where no maintained connector covers your case.
Synchronous checks on cart and order, and event-driven sync where near-real-time is enough.
Stock, pricing, and order flows synchronized with the ERP your finance team already operates.
Payment providers connected with capture, refund, and reconciliation flows tested on real orders.
Product data, fulfilment, and content joined to commercetools with one owner per field agreed.
Retry logic, failure alerting, and version upkeep once the connections are carrying live traffic.
Not sure which of your systems need custom work?
Ready connectors already exist across most of these categories, and our commercetools integrations directory lists them by vendor. Whether one fits depends on whether your instance uses the standard fields it expects.
SAP, NetSuite, Microsoft Dynamics 365, and Oracle ERP among others, connected for stock and order write-back.
Manhattan, Blue Yonder, ShipStation, and carrier APIs joined to fulfilment and returns flows.
Algolia, Bloomreach, Constructor, and Klevu fed from the catalog so ranking rules survive a reindex.
Almost every overrun we are called into traces back to the same omission: nobody decided which system owns each field before the build started. When the ERP and the PIM both claim price, the connector works and the data is still wrong.
The same architects handle the storefront calling these APIs, so nothing is thrown over a wall: see our headless commerce development work.
One named owner agreed for every synchronized field
Reconciliation rules written before any connector is deployed
Failure and retry behavior specified, never left to defaults
Every flow tested against real order volumes before launch
Every project opens with a data map rather than a connector. We write down which system owns each field, what happens when two disagree, and how a failed message is retried. The build follows that document.
We list every system in scope, the fields each one owns, and the direction each flow travels. The finished map is yours to keep, even if the build never happens.
Each connection is classified against the four patterns, starting from what already exists in our connector library. That split is what sets the estimate.
Standard connectors are configured and deployed on commercetools Connect. Custom services are written against the APIs and deployed alongside them.
Every flow is exercised with production-scale data, including the failure paths, so retry and alerting behavior is proven before customers depend on it.
Alerting is configured on each flow and the connections stay under support, including the version upkeep that commercetools releases and vendor API changes require.

23+
Years in eCommerce
600+
In-house specialists
$4B+
Processed through client stores yearly
%20(1).png)
Most estates are a mix, and the choice is per flow rather than per system. The moment your instance carries custom fields or your reconciliation rules are specific to your business, a written service is cheaper than fighting a connector's assumptions.
Browse what already exists in the connector directory.
An API extension is a synchronous call commercetools makes out to your service while a cart or order is being written, so the response can change the result before it is saved. Tax and fraud checks are the common cases. It is the right pattern when the answer has to be correct at that instant, and the wrong one when a few seconds of delay would be acceptable.
Somebody has already decided, usually by accident. Until one system is named as the owner of each field, both write and the last one wins, which is why the symptom shows up as intermittent wrong prices rather than a failure. We name an owner per field in the data map, then make every other system read that value rather than compete with it.
Against production-scale data, including the paths that fail. We exercise each flow at real order volumes, then deliberately break it: time out the receiving system, send a malformed payload, replay a duplicate event. Retry and alerting behavior is proven before customers depend on it, because integration defects surface as wrong data rather than error pages.
The count of custom services sets the number, not the count of systems. An estate served by ready connectors costs a fraction of one needing three bespoke services with their own reconciliation rules. We price after the systems audit, never before it. Vendor connector licenses and your commercetools plan are billed to you directly and are separate from our invoice.
Latency and ownership decide it. If the answer must land before the order is saved, it is an API extension. If the receiving system only needs to know shortly afterwards, a subscription carries the event without holding up checkout. A custom microservice earns its place when the logic is yours rather than either system's, such as reconciliation rules no vendor connector models.
Often, and it is usually cheaper than replacing it. Many teams fork a vendor connector to add one field and then cannot take upstream updates. We look at whether your changes can move into an API extension or a subscription alongside the stock connector, which puts you back on the vendor's release path and keeps your logic separate from theirs.
Yes. We start by reading what exists: which flows are live, which are silently failing, and how much lives in forked connector code outside version control. You get a written assessment naming what to keep, what to rewrite, and what to retire, with an estimate against each item. That assessment is yours whether or not we do the work.
The engineers who wrote them. Alerting is configured per flow, and support covers the two things that break working integrations over time: commercetools platform releases, and vendor API changes at the other end. Both arrive on someone else's schedule, so upkeep is a standing commitment rather than a warranty period.
Name what you operate today and which pieces already exchange data. We come back with each connection classified as standard, configured, or custom, and an estimate against the custom ones.
Prefer to talk now? Book a call straight away, or email us at: [email protected]
An official commercetools partner, with the data map written before the first connector is deployed.