BPO, GBS & shared services

Because the denominator is enormous. Five percent of a transaction volume in the hundreds of thousands is a steady stream of defects reaching clients — and each one is a potential SLA breach, escalation, or credit that erodes the contract at renewal. A rate that reads as healthy in a low-volume business is alarming at scale, because the aggregate is what the percentage hides. The honest way to see it is as absolute volume: not "we are at 95%" but "this many imperfect deliveries a month, concentrated in these clients." Expressed that way, the case for closing the gap is obvious.

By defining a standard precise enough that a defect means the same thing on every account. In most multi-client operations each relationship marks its own homework — one team strict, another lenient — so the scores are not comparable and leadership cannot tell whether a low number reflects worse work or harder marking. Defining quality once, explicitly, in terms that travel across clients is the foundational work; it is what makes every downstream number honest. It is unglamorous, but it is the difference between a dashboard people trust and one they quietly argue with, and it is the precondition for governing the book as a whole.

Because the quality systems that held at normal volume were never stress-tested for the peak, and they break exactly when a surge or a short-staffed shift raises the stakes. The defence is leading indicators — signals that move before the SLA does, like queue depth climbing, first-pass yield dipping, or handling time creeping up. These turn hours or days before a breach and give an operations leader the chance to act while the SLA is still intact. A governance system built on lagging numbers only tells you that you missed; one built on leading signals lets you avoid missing in the first place.

Not by asking people to be more careful. At that level the residual defects are not random lapses — they cluster into a handful of systemic failure modes the current process reliably produces. The work is to segment those remaining defects, find the recurring root of each, and design that cause out of the operating model rather than inspecting harder for it. This is exactly how an operation moved from 95% to 99% across more than 2,000 campaigns and 450 clients. The last few points are the hardest because they are systemic, and the most durable, because once the cause is removed the defect does not return.

With a BI layer that reconciles across every account, so the same defect recurring across many clients appears as one pattern rather than a scatter of unrelated incidents. The most dangerous blind spot in shared services is the gap between accounts: when quality is reported per client, a systemic defect hides in plain sight because no single account owner sees enough of it to raise the alarm. Reconciled reporting across the book turns that scatter into a visible problem with one root cause — which lets you fix a systemic failure once instead of forty times, and catch it before it becomes a wave of escalations.

It holds only if it is governed; otherwise it decays. The energy of a clean-up fades, attention moves to the next fire, and within a couple of quarters the score drifts back and the same problems resurface at the next audit. What prevents that is a governance cadence: a small set of trusted quality metrics on the operating review, a named owner accountable for the number across the book, thresholds for when something needs attention, and clear decision rights for who acts when a client trends the wrong way. Most operations skip this half — they build reporting and call it governance. The cadence is what makes the gain permanent.

Yes. The core challenge — consistency at scale across many internal clients or business units — is the same in a captive global capability centre as in a third-party provider, and so is the solution: a precise standard, reconciled measurement across the book, and a governance cadence that holds the gain. If anything, a captive centre often has cleaner access to the upstream processes that generate defects, which makes designing out systemic failure modes more tractable. The principles travel across the operating model whether the clients are external customers or internal functions being served by shared services.

More analysts give you more inspection; they do not give you a system. You can add headcount to checking and still see the same defects recur, because the missing pieces are a consistent standard, reconciled visibility across the book, and a governance cadence that turns findings into action and designs systemic causes out. Inspection catches errors after they happen; a quality function stops them being produced. Done well, this work often makes your existing QA team far more effective, because for the first time their findings are comparable across clients and wired into a rhythm that actually removes the recurring causes rather than re-finding them every cycle.