Offshore, onshore or hybrid: shaping a Kinaxis delivery team that actually ships

Every Kinaxis programme eventually becomes a conversation about where the team sits. The conversation usually arrives as a cost question. It is actually a design question, and the answer shapes what the programme can deliver.

Three failure modes

All onshore. Senior, responsive, expensive — and thin. When build volume peaks, the same architects who should be reviewing design are writing workbook logic at midnight. Quality holds until it suddenly doesn’t.

All offshore. Cost-efficient and tireless, but the design conversation with planners happens through a specification, and specifications lose the argument every time. The team builds what was written, the planners reject what was built, and both are right.

Hybrid without a seam. The most common shape and the most common failure: onshore and offshore teams working from the same backlog with no defined review loop. Nobody owns the join, so nobody notices when the two halves drift.

The split that ships

The model we run — and the one we’ve watched succeed inside the large integrators — puts each activity where it belongs.

• Design sits with the planners. A small, senior onshore team owns the solution architecture, the data model and the planning process decisions. They are in the room, or on the same time zone as the room, because design is a negotiation and negotiations need presence.
• Build scales where capacity is. Workbook, analytics and interface build runs with an offshore practice working against a signed design, in sprints, with demo at the end of every one.
• Review is the seam, and it is owned. The architects who wrote the design review every sprint’s output before the planners see it. Not a gate at the end; a loop every fortnight.
• Adoption is local. Training, go-live support and the first three planning cycles happen beside the planners. Knowledge transfer is a sprint deliverable, not a closing workshop.

Three questions for any team proposal

1. Who owns the join between design and build, by name?
2. What happens to a sprint that fails review — does it ship anyway to hold the date?
3. How many of the on-site people will still be there for the third planning cycle after go-live?

A proposal that answers all three has a team shape. A proposal that answers only the rate card has a cost.

Our Three Gates Framework makes the seam explicit: nothing is configured until the design is signed at Gate 2, and every Accelerate sprint includes a review and a demo. We shape the team around that sequence — UK/EU architects, an APAC delivery practice, and planners in every demo.

Designing the seam

The join between design and build is where hybrid teams succeed or fail, and it is almost never specified. Four things make it work.

A design that is actually finished. Not a slide pack — a signed data model, planning hierarchy, integration pattern and workbook specification, with the exceptions written down. Offshore build against an unfinished design converts capacity into rework at full speed.

A named reviewer per sprint. The architect who wrote the design reviews the output before the planners see it. Named, not a function; the same person, not a rota. Rotating reviewers produce inconsistent decisions, and inconsistency in a data model compounds.

Overlap hours that are real. Two to three hours of genuine overlap, protected in both calendars, is enough. Zero overlap turns every clarification into a 24-hour round trip, and a build team that waits a day for an answer will guess instead.

A path for the build team to challenge the design. The people writing the workbook logic find the design’s flaws first. If the only channel is a defect ticket, they will build the flaw as specified and log it. Give them a named person to ring.

What each location should own

• Onshore, with the planners: solution architecture, data model, planning hierarchy, the S&OP process decisions, and the first three planning cycles after go-live.
• Offshore, against a signed design: workbook and analytics build, interface development, test execution, data migration scripting, and sustained post-go-live support.
• Either, explicitly assigned: integration architecture — it needs both the SAP side and the Kinaxis side in the same head, and where that head sits matters less than that there is only one of them.
• Never split: ownership of the data model. One owner, wherever they sit.

The economics, without the sales gloss

A hybrid does not halve your cost. Realistically it takes 20–35% off a comparable all-onshore team once you account for the review overhead, the overlap hours and a slower first sprint while the build team learns the estate.

What it buys that matters more than the saving is elasticity. Build volume on a Kinaxis programme is spiky — it peaks during workbook construction and again during test. An offshore practice can absorb that peak without you carrying the bench through the troughs.

If the proposal in front of you claims a 50% saving with no review overhead, the review overhead has been deleted rather than costed, and you will meet it later as defects.

Three questions that reveal the truth

1. Who owns the join, by name? A firm with a real model answers with a person in under five seconds.
2. What happens to a sprint that fails review? “It ships anyway to hold the date” is the honest answer most of the time. If they say that, at least you know what you’re buying.
3. How many of the on-site people will still be there for the third planning cycle after go-live? The answer is the difference between a delivery team and a launch team.

The shape we run

UK and EU architects own design and sit with your planners. An APAC practice builds at volume against the signed design, in sprints, with the same architects reviewing every one. Adoption happens locally, and knowledge transfer is a sprint deliverable rather than a closing workshop.

It is not the cheapest shape available. It is the one where the person who decided how your planning works is still reachable in month nine — which is the only thing that has ever predicted whether a programme still runs a year after go-live.