The choice is usually presented as binary. Keep the operation in Britain, or move it offshore. Argue about accents, argue about cost, pick a side.
Most well-run UK service operations stopped answering the question that way some years ago. They run both, deliberately, with work allocated between them on a principle rather than by historical accident — and the principle is almost never about complexity.
Sort by accountability, not difficulty
Not difficulty. Consequence, and specifically whether an interaction requires somebody who can commit the organisation.
Work that prepares, investigates, documents, services and resolves within defined limits travels perfectly well. Work that constitutes a determination — a final complaint outcome, a credit decision, a decision to end a customer relationship, anything a regulator expects a named individual to own — does not travel, and no commercial argument has ever successfully moved it.
The useful consequence of this framing is that it cuts across difficulty. Some of the most demanding conversations an operation handles can sit offshore, because the agent is gathering and resolving within authority — which is a large part of what British firms are actually buying when they contract outsourced call centre capability in the Philippines. Some comparatively simple decisions must stay onshore, because of who has to own them.
Organisations that sort by complexity instead produce the characteristic blended-model failure: a difficult but perfectly transferable queue kept in Britain at four times the cost, and a simple-looking decision quietly made offshore that somebody will later have to explain.
“Everyone sorts by how hard it looks, and that is the wrong axis entirely,” said Ralf Ellspermann, founder and chief strategy officer at PITON-Global. “Sort by who has to be accountable for the answer. The rest follows from that and you stop having the argument every six months.”
Where the onshore team earns its cost
A small retained British team is not a concession to sentiment, and advisers who design blended UK and offshore operations tend to spend more time sizing it than they spend on the provider shortlist. It does four things that genuinely require proximity.
Determination
Everything above, concentrated in a group small enough to be consistent.
Escalation of last resort
The small number of cases that have stopped being operational and become reputational — the ones that will reach a chief executive’s inbox, a journalist, or the Financial Ombudsman Service. These need somebody who can act without asking.
Ownership of the offshore relationship
This is a real job requiring a real person: performance, scope, quality calibration, and the authority to say no to a change. Programmes where nobody onshore owns the provider relationship degrade predictably, and the degradation is slow enough that it is usually noticed a quarter too late.
The internal interface
Legal, risk, compliance, the product teams and the complaints function all expect a colleague who sits in the same building and attends the same meetings. An offshore operations manager, however capable, cannot do this, and an operation without it ends up with policy changes it finds out about after they have taken effect.
That is typically a handful of people rather than a floor, and sizing it honestly is the hardest conversation in the design, because the instinct is to retain more than the function requires.
“Every client asks how many people to keep in Britain and almost every one of them keeps too many,” said John Maczynski, chief executive of PITON-Global. “Count the decisions that need a named owner. That is the number, and it is smaller than the structure they arrived with.”
Where blended models actually break
The join between the two teams is where blended models break, and the failure is always the same: a customer who has to explain themselves twice.
An interaction escalating from offshore to onshore carries context. If that context does not travel — because the systems differ, because the note was thin, because the handover is a queue rather than a conversation — the customer repeats themselves at precisely the moment they are already frustrated.
“The seam is where blended models die, and it dies from thin notes,” said John Maczynski, chief executive of PITON-Global. “Nobody writes a bad note on purpose. They write one because the handover standard was never specified and nobody reviews handovers.”
The seam holds when three things are in place. A single case record both teams work in, rather than two systems with a bridge. A defined escalation standard stating what information must accompany a transfer, enforced in quality review. And warm handover for anything sensitive, meaning a conversation between the two agents rather than a ticket.
What automation did to the balance
The blended question used to be about volume. It is increasingly about judgement density.
A great deal of the simple traffic that once filled an offshore floor is now resolved before a person sees it. What remains is harder, which has two effects pulling in opposite directions.
It raises the skill requirement offshore, which argues for a more senior and more expensive offshore team than the original business case assumed. And it reduces the total human volume, which makes a small retained onshore team proportionally more significant in both cost and influence.
Organisations that have not revisited their allocation since automation arrived are generally carrying the wrong shape — an offshore team sized for a volume that has gone, and an onshore team sized for a judgement load that has grown.
Sequencing a move to blended
For an organisation starting from wholly onshore delivery, the order matters more than the destination.
Start where volume is high and consequence is low, demonstrate the controls hold, and widen scope on the strength of results rather than projections. Determination stays in Britain the whole way through. Settle the authority schedule before go-live, not in the aftermath of the first incident that exposes its absence.
And resist redesigning the process at the same time as relocating it. Changing both simultaneously makes it impossible to attribute any subsequent problem to either, which is the diagnostic position nobody wants when a service level starts slipping.
Provider selection gets the attention. Sequencing decides the outcome, and the order of the move is where most of the avoidable damage is done.
Harder to run, and worth it
Blended delivery is more complicated to run than either pure model. It requires two management relationships, a defined seam, and somebody senior who owns the whole.
What it buys is the ability to put each piece of work where it belongs — scale and coverage where those are what is needed, accountability and relationship where those are. For most UK organisations of any size, that allocation produces a better operation than either alternative, which is why the binary framing has largely disappeared from serious conversations even though it survives almost intact in the trade press.
