Four models get compared as if they were competing products. They are not. They are four points on a single axis: how much delivery management you want to own.
Get that axis right and the choice is usually obvious. Get it wrong and you end up with either three separate reporting lines into a manager who did not ask for them, or a captive entity built for a capability that turned out to be temporary.
A note on the acronym
GCC here means Global Capability Centre — a captive offshore entity owned by the parent company. In a regional context the same three letters mean Gulf Cooperation Council, and the two get confused often enough in vendor material to be worth stating once.
The four models
| Who directs the work | Typical size | Who carries hiring and attrition | End state | |
|---|---|---|---|---|
| Staff augmentation | You | 1–5 | Vendor | People leave when the gap closes |
| Dedicated pod | Vendor's delivery lead | 4–10 | Vendor | Pod scales or stops |
| Offshore delivery centre | Vendor, against your roadmap | 10–50+ | Vendor | Standing unit, indefinite |
| BOT → GCC | Vendor, then you | 20+ | Vendor, then you | Your own legal entity |
Staff augmentation
Use it when the gap is one or two specific skills and your process already works. A data engineer for a migration. An SRE for a scaling problem. A mobile lead for two quarters.
Where it stops working: above about five people. Each additional person is another reporting line into your managers, and the management load you did not price grows faster than the capacity you gained. Teams discover this when their engineering lead spends more time coordinating contractors than engineering.
The other failure mode is rotation. If the supplier can swap people without your agreement, the domain knowledge you paid to build leaves with them.
Dedicated pods
Use it when the work is a continuous stream and you would rather review outcomes than assign tasks. The pod arrives with its own delivery lead, so accountability is for output rather than for hours.
Where it stops working: when the roadmap needs more than about ten people, or when the pod's work spans so many of your systems that a shared context with your own team becomes the bottleneck.
Offshore delivery centre
Use it when it is a programme rather than a project, and you want a standing unit with its own hiring pipeline, its own retention problem and one accountability line.
Where it stops working: below scale it is overhead you are paying for nothing — an ODC is worth its structure above roughly fifteen engineers, and below that two pods usually cost less and carry the same accountability. At the other end, if the capability is genuinely core and permanent, you are renting something you should own.
The question nobody asks early enough
What happens to the knowledge if the contract ends? An ODC that has run for three years holds a great deal of understanding about your systems. Whether that is an asset or a hostage depends on documentation and cross-staffing decisions made at the start, not at the exit.
Build-Operate-Transfer, and the GCC it becomes
Use it when the intent is a captive entity and the question is only how to get there. The vendor incorporates, hires, runs payroll and compliance, and operates the unit — then transfers it to you on a date agreed at signature.
The clause that decides whether it is real BOT: transfer date, price and mechanism fixed in the first contract. Where those are left to be negotiated later, it is outsourcing with a friendlier name, and you will discover the price at the moment you have least leverage. Ask to see that clause before signing, not when you want to use it.
Where it stops working: when the capability turns out to be temporary. An entity is expensive to unwind — statutory closure in India is slower than incorporation — so premature BOT converts flexibility into a fixed cost.
Choosing, in four questions
- How long? Under a year, augmentation or a pod. Multi-year, ODC or BOT.
- Who manages? If you have delivery management capacity and want control, augmentation. If you would rather buy the management, everything else.
- Permanent capability? If yes, BOT has a destination the others do not.
- What scale? Under five people, augmentation. Five to fifteen, a pod. Above fifteen, an ODC starts to justify itself. Above roughly fifty and permanent, a GCC's fixed cost amortises.
The mistake that costs most
Starting at the wrong end of the axis. Two patterns recur:
Scaling augmentation past its limit because it started well. Fifteen individually-managed contractors is not an offshore team; it is fifteen coordination problems, and the people carrying them are your own leads.
Building an entity too early because the strategy deck said captive. If the roadmap that justified it is cancelled, the entity remains — with a lease, a payroll and a statutory calendar.
Moving up the axis is straightforward. Moving down it is not, which is an argument for starting one step lower than the plan assumes.