Employee experience in shared services has spent a decade being treated as an HR programme — engagement surveys, recognition schemes, a table tennis table on the operations floor. The centres with the lowest attrition rarely have better programmes. They have better work design. That distinction matters more this year than it has in a decade. Delivery locations are under extraordinary human strain, hiring is restarting in the markets that shut down last year, and the people who carry undocumented process knowledge have discovered they can be interviewed from their living rooms. Retention in a shared services centre is an operating model problem wearing an HR badge, and the organisations that keep treating it as a morale exercise will spend the next eighteen months rebuilding teams.
Start with the shape of the attrition curve, not the rate
A single annual attrition percentage tells you almost nothing. Three derived numbers tell you nearly everything, because each one indicts a different system. Attrition in the first ninety days is a hiring and onboarding failure. People left because the job was not the job described, or because nobody taught them how to do it and they were too embarrassed to keep asking. This is the cheapest attrition to fix and the most consistently ignored, because it is recorded as a bad hire rather than as a bad process. Attrition between twelve and twenty-four months is a career design failure. These are people who became competent, looked upward, saw one role — team leader — and concluded correctly that the queue was long. Losing them is expensive because they had just crossed from cost to contribution. Variance in attrition between comparable teams is a management failure, and it is the most under-examined statistic in the industry. When two teams run the same process, on the same systems, with the same shift pattern and the same pay bands, and one loses eight per cent a year while the other loses thirty, the difference is the person running the team. Everyone in the centre already knows which teams those are. Publish those three numbers monthly. They will reorganise the conversation faster than any survey.
Your metrics are the employee experience
Whatever you count becomes what the day feels like. A centre that manages utilisation converts every idle minute into a moral failing, and teaches people to look busy. A centre that manages average handling time teaches people to close tickets, not to solve problems, and quietly punishes the person who stops to fix a root cause. Strict adherence scheduling turns a bathroom break, a prayer, or a call from a child's school into a compliance event. None of this is written in a policy. All of it is felt daily, and it is the reason people describe the work as relentless rather than difficult. The alternative is not softer targets; it is different ones. Measure exception ageing, quality escapes, rework rates, and the volume of work eliminated — outcomes that reward the person who improves the process rather than the person who processes fastest. Give teams authority over a defined band of decisions so that judgement exists somewhere in the role. Autonomy inside a narrow band beats enrichment programmes outside the work.
The tower trap
Functional towers produce a single promotion path: processor, senior processor, team leader. The consequence is an industry-wide habit of taking excellent practitioners and making them mediocre managers, thereby losing both the practitioner and, eventually, the manager. Build at least three ladders besides the management one, each with real pay bands: Process specialist, owning the design and documentation of a process end to end. Automation and data analyst, owning capture, matching and exception reduction. Controls and compliance specialist, owning evidence and audit interaction. A fourth, service manager, suits people who want client contact without hierarchy. The second half of career design is pricing scarcity rather than seniority. Statutory knowledge for a specific jurisdiction, a hard-to-hire systems skill, the ability to handle a regulator or an auditor in the local language — these should command premiums regardless of tenure. Most centres instead pay a flat band and then lose exactly the capability they cannot replace.
Ask before they leave
Exit interviews are archaeology. People are polite, they name compensation because it is the socially safe answer, and the information arrives after the decision. Stay conversations at month nine and month eighteen — conducted by someone other than the direct manager, with a small number of specific questions and a visible action from the last round — produce different answers: the shift pattern, the commute, the manager, the feeling of being a resource number in a capacity model. Compensation matters, but it is rarely the trigger; it is the justification people use for a decision made about respect and progression.
Practical Guidance for Retention Strategy Review
- Report the three attrition curves, not the headline rate. Ninety-day, twelve-to-twenty-four-month, and inter-team variance.
- Audit your operational metrics for what they teach. Any metric that penalises fixing the root cause is manufacturing attrition.
- Cap spans of control at a number where a manager can actually know their team, and remove administrative load from team leaders so managing is the job.
- Build specialist ladders with published pay bands, so that competence has somewhere to go other than management.
- Pay premiums for scarce capability, not for tenure, and be explicit about which capabilities qualify.
- Run stay conversations at nine and eighteen months and publish what changed as a result.
- Treat shift patterns and schedule predictability as compensation, because employees do.
- Decide deliberately whether each process is designed for churn or for retention, and fund it accordingly.
The Regional Angle
Retention arithmetic works differently in the Gulf, and the first difference invalidates the headline metric entirely. When residency is tied to sponsorship, leaving an employer is not a decision about a job. It is a decision about a visa, a notice period, dependants' residency, a medical, school enrolment, a tenancy contract usually paid a year in advance, and in some cases a period without legal status while paperwork moves. The friction is enormous, and it suppresses voluntary exits far below the level of actual disengagement. A centre in this region can therefore report attrition of eight per cent and be in serious trouble, because the people who have mentally resigned are still at their desks, performing to the letter of the job description and nothing beyond it. Two consequences follow. Attrition is a lagging and distorted retention indicator here, so measure the leading ones instead — internal transfer requests, participation in optional work, absence patterns, stay conversation themes. And understand that suppressed attrition behaves like a dam: when a large employer opens a comparable centre in the same city and takes on the visa transfer cost, the release is sudden and concentrated in exactly the people you could least afford to lose. The second difference is that the compensation conversation is not primarily about salary. For a large share of the workforce, the decisive line items are the education allowance, the annual flight, housing, and medical cover that extends to dependants. A salary increase that fails to cover a school fee increase is experienced as a pay cut, correctly. This also gives the resignation cycle a calendar: decisions cluster around the end of the academic year and the long consolidated summer leave, when families reassess. Retention interventions that land in September are addressing a decision taken in April. Budget and act against that rhythm, and treat allowances as a strategic instrument rather than as legacy policy nobody has revisited since the centre opened. The third is the emerging competition for national talent. With Saudi Arabia pressing companies to establish regional headquarters in the Kingdom to remain eligible for government contracts, and with nationalisation targets applying across the region, a limited pool of experienced national finance and operations professionals is being pursued by every employer at once. Pay alone will not hold that group, because pay can always be matched; career design, genuine scope, sponsorship for professional qualifications and visible progression will. Centres that treat nationalisation as a compliance quota fill seats and lose them annually. Centres that treat it as a talent programme build a cohort that becomes their management layer.
The human situation this month
It would be dishonest to write about employee experience in shared services in May 2021 without naming what is happening in the largest delivery market. Colleagues in Indian centres are living through a devastating wave: people are losing parents and partners, many are ill themselves, and hospital beds, oxygen and vaccination appointments have been the substance of daily life for weeks. Centres in South East Asia have spent much of the last two months under renewed movement restrictions. How employers behave right now will be remembered for a decade, by the people who stayed and by everyone they talk to. The providers acting well are running vaccination programmes for staff and families, funding medical support and hospitalisation help, guaranteeing pay through illness and bereavement, extending leave without a process, and — quietly — telling clients that some service levels will not be met this month. For buyers, there is one obligation: do not spend this period enforcing service credits. Ask what help is needed, agree temporary priorities in writing, and record the accommodation. The commercial value of that behaviour will exceed anything the contract could have recovered, and the relationship consequences of the opposite choice will outlast the pandemic.
The objection worth taking seriously
The objection is straightforward: employee experience is a phrase that appears when labour markets tighten and vanishes when they loosen. Shared services exist to deliver work at a lower unit cost, every recommendation above raises cost, and for genuinely standardised high-volume work a churn model can be entirely rational — hire continuously, train quickly, document thoroughly, accept thirty per cent annual attrition and price it in. Plenty of centres have run that model profitably for years. That is a real strategy, and it deserves to be stated rather than denied. The trouble is that almost nobody runs it deliberately. The churn model works only when the processes are genuinely standardised, the documentation is genuinely current, the systems genuinely prevent errors, and the training pipeline genuinely produces competence in weeks. Most centres claim those four conditions and satisfy none of them, which leaves them running churn economics over retention-dependent processes — and paying for it in errors, escalations, rework and client trust rather than in a visible cost line. So make the choice explicit, process by process. Design for churn and fund the documentation, automation and controls that make churn survivable. Or design for retention and fund the career structures, management capacity and compensation that make retention real. The failure mode is the default: a churn cost base carrying a retention risk profile, discovered only when three experienced people leave in the same quarter.
Common Questions
What is a good attrition rate for a shared services centre?
The wrong question. Compare your ninety-day rate, your twelve-to-twenty-four-month rate and your inter-team variance against your own trend and against the criticality of the roles involved. A centre at twenty per cent with stable specialists is healthier than one at twelve per cent that loses its process owners.
Does remote work help retention in delivery centres?
It has become one of the strongest instruments available, particularly for staff who relocated to home towns last year and will not return to metropolitan rents for the same salary. It also removes a commute that was, for many, the worst part of the job.
How much of this is the client's responsibility?
More than most clients accept. Buyers who demand named individuals, refuse cross-training, escalate constantly and price on full-time equivalents are actively shaping the employee experience inside their provider — and then complain about turnover.
What should we expect over the next twelve months?
Expect attrition to rise sharply in the second half as hiring restarts and vaccinated markets reopen, with the steepest increases among two-to-five-year staff in technology-adjacent finance roles. Expect location flexibility to become a permanent, priced element of the package in delivery geographies rather than a pandemic concession. Expect buyers to start requiring attrition and knowledge-continuity reporting in governance meetings, and expect the better providers to offer it unprompted. And expect employer conduct during this wave to become a hiring differentiator that no recruitment campaign will be able to buy back.
Retention Strategy Review — we examine your attrition curves, operational metrics and career structures together, because in shared services the people problem is almost always an operating model problem first.