The labour figures published this morning put the number of Americans who quit their jobs in August at a record 4.3 million. The phrase attached to that number has been circulating since the spring, and it will now be quoted in every board pack for the rest of the year. Outsourcing floors got there first. Voluntary attrition in offshore delivery has never been low — contact centre operations in the Philippines have run in the low thirties annually for years, and Indian transaction processing is not far behind — but providers this year are reporting materially worse, with the sharpest deterioration among staff in their first year. The number nobody publishes is the ninety-day failure rate for new hires, and it is the one that explains what clients are actually experiencing. Because here is the thing that catches buyers out: your service levels are probably still green.
What you actually bought
You did not outsource a process. You outsourced the small group of people who know the exceptions. Every back-office process has a documented flow and a real one. The documented flow says that invoices arrive with purchase orders and are matched automatically. The real one includes the supplier who has sent unreferenced invoices for nine years, the customer whose remittance advice arrives as a scanned Arabic attachment, the intercompany recharge that posts to a suspense account until somebody in the group recognises it, and the fourteen other cases that constitute most of the effort. That knowledge is not in the process map. It sits with the two or three people on the account who have been there longest. When those people leave, the process does not stop. It degrades in a way your provider's metrics are not designed to detect. Volumes are still processed inside the agreed window, quality sampling still passes, and the monthly service review still shows a wall of green. Meanwhile the number of queries arriving at your finance team doubles, exceptions take longer to resolve, and your own staff spend their evenings answering questions the account used to answer for itself. The cost of provider attrition transfers to the client, and it transfers in a line item nobody tracks.
The five leading indicators
Before the service level breaks, these move. Track them from your side, monthly. Query volume from the provider to your retained team, normalised against transaction count. Escalation rate on a stable transaction mix. The proportion of items classified as exceptions, which should be flat or falling and quietly rises when confidence drops. Time to first touch on non-standard items rather than average handling time, because averages are dominated by the easy majority. And at close, the number of prior-period adjustments and correcting journals, which is the clearest single signal that judgement has left the account. If three of those five have deteriorated over two quarters while your dashboard is green, you have an attrition problem that the contract is not measuring.
| Indicator | What the client watches |
|---|---|
| Provider queries | Query volume relative to transaction count |
| Escalations | Escalation rate on a stable transaction mix |
| Exceptions | Share of items classified as exceptions |
| Non-standard items | Time to first touch, rather than a blended average |
| Close corrections | Prior-period adjustments and correcting journals |
Qualitative summary of this article's source text, not a measured outcome or performance estimate.
Ask for the numbers you are not being given
Most outsourcing agreements report attrition at company level, annually, if at all. That figure is useless. What matters is your account. Ask for monthly attrition at account level, split into voluntary and involuntary, and broken down by tenure band. Ask for the proportion of people working on your account with less than six months on it — this is the single most predictive number available and almost no client requests it. Ask how long vacancies stay open, because holding positions unfilled is a standard margin protection technique and it is your service that absorbs it. Ask for continuity commitments on the roles that matter: team leads, process specialists, and whoever holds the exception knowledge on each tower. Providers resist this on the grounds that staffing is their business. The reasonable answer is that staffing composition is an input you are paying for, and you are asking for visibility rather than control.
Practical Guidance for Back Office Talent Strategy
- Require account-level attrition reporting monthly, split by voluntary, involuntary and tenure.
- Set an experience-mix service level — a cap on the share of account staff below ninety days.
- Make knowledge an owned asset: exception registers and decision rules maintained by the provider, reviewed quarterly, and held in your repository.
- Do not pay full rate for training, and define when a new joiner becomes billable.
- Require shadow coverage on named critical roles rather than a general commitment to resilience.
- Track the five client-side indicators and put them on the monthly governance agenda beside the service levels.
- Reduce the exception load, because every exception you eliminate removes a dependency on somebody's memory.
- Keep judgement work in a retained team, and be explicit about which decisions never leave.
The price conversation is coming, and refusing it is expensive
Wage inflation in delivery locations is real, and providers are asking for mid-term increases. The instinct is to refuse, and refusal is usually the expensive option: a provider with a margin problem and no price relief manages it through staffing, which means juniorisation, unfilled vacancies and the exact degradation described above, arriving as your problem rather than theirs. The constructive trade is specific. Accept a modest, indexed increase; in exchange, require attrition transparency, an experience-mix commitment, documented exception registers, and a productivity path that lets the provider absorb future cost through automation rather than through thinner staffing. That conversation produces a better contract than either an unexamined increase or a flat refusal, and it can be had this quarter while there is still goodwill on both sides.
The Regional Angle
Three things make this materially different for a business operating from the Gulf, and the first will invalidate a transition plan. When clients change outsourcing providers elsewhere, the standard assumption is that the incoming provider rehires most of the outgoing team, so process knowledge stays on the account even as the logo changes. That assumption is unreliable here. Where delivery is onshore, staff are employed on sponsored residency tied to a specific employer, and the transfer of a person from one provider to another involves visa cancellation and reissue, no-objection practice, notice periods that may be worked or bought out, and sometimes a gap during which the individual cannot legally work. Some will not wait; many will simply leave the country. The result is that a provider change in this region loses far more institutional knowledge than the same change in London or Manila, and the transition plan that assumes rebadging is quietly assuming something that will not happen. Build the transition around documented process knowledge and a parallel run, and price the shadow period properly rather than treating it as a contingency. The second is a concentration problem hiding inside an apparent abundance. A great deal of Gulf back-office work has been offshored to India, the Philippines and Egypt, and for English-language transaction processing the labour pool is deep even at high attrition. Arabic-language work is not. The people who can read a handwritten Arabic delivery note, handle a bilingual statutory filing, correspond with a ministry portal or reconcile an Arabic bank statement are a small fraction of any offshore centre, and they are frequently one or two individuals per account. Attrition of two per cent in the English team is a staffing event. Attrition of one person in the Arabic team is an outage. Identify those roles explicitly, name them in the contract, require a trained backup for each, and stop reporting attrition as a single blended percentage across an account where one segment is ten times more fragile than the other. The third is what is actually triggering resignations in regional teams this quarter, and it is not a global cultural shift. Salary increments were frozen across most of the region in 2020, and they are being unfrozen now — unevenly, and against a market that has moved. The consequence is that new hires are arriving on packages above those of incumbents doing the same job, in a compact expatriate labour market where pay information travels through community networks faster than through any human resources department. Internal pay compression, not the discovery of remote work, is what makes a five-year employee update a curriculum vitae. Run the internal equity review before the retention scheme, because a retention bonus paid to somebody who has just learned what the new joiner earns solves nothing.
The objection worth taking seriously
The strongest objection is straightforward: attrition is the provider's problem. That is the point of outsourcing. You buy a defined outcome at a defined price, and how the supplier staffs it is precisely the risk you transferred. A client demanding tenure data, experience mixes and named individuals is re-insourcing management responsibility while continuing to pay somebody else to hold it, and every clause of that kind raises the price. Attrition figures are also easy to game — move people between accounts, reclassify a resignation as a redeployment — so the transparency being demanded may not even be real. Most of that is fair, and clients who try to run a provider's human resources function deserve the bill they receive. But the risk transfer is narrower than the language suggests. What was transferred is the measurable portion of the process: volumes, cycle times, accuracy on sampled transactions. What was never transferred is the unmeasured portion — exceptions, judgement, escalation, and the institutional memory of why this customer is treated differently. That part degrades silently and lands on your own cost line, which is why a client can simultaneously report full service level attainment and watch its retained finance team work later every month. The remedy is not to manage the provider's staffing. It is to measure the things that actually deteriorate, and to make process knowledge an asset you own rather than a favour your supplier's longest-serving employee does for you. Do that, and provider attrition becomes a supplier problem again — which is what you paid for.
Common Questions
What attrition rate should we treat as a red flag?
The absolute number matters less than the trend and the tenure mix. An account where a third of staff have under six months on the process is in trouble regardless of the headline rate.
Should we agree to a mid-term price increase?
Usually yes, if it buys transparency, an experience-mix commitment and a documented knowledge obligation. Refusing outright tends to purchase juniorisation at the old price.
Can we require named individuals to stay?
You can require continuity commitments, notice of changes and trained backups for named roles. You cannot prevent resignations, and a clause that pretends otherwise will be quietly ignored.
What should we expect over the next twelve months?
Expect attrition in delivery locations to stay elevated well into next year, with wage inflation following it, and expect provider price increase requests to become routine rather than exceptional at renewal. Expect providers to respond with automation of the highest-volume, lowest-judgement work, which will improve the economics and concentrate the remaining human work in exactly the exception-heavy areas where attrition hurts most. Expect experience-mix and knowledge-retention clauses to start appearing in mainstream outsourcing templates, because buyers who were burned this year are drafting them now. And expect the quiet winners to be the organisations that spent this period eliminating exceptions rather than negotiating rates, because they will need fewer irreplaceable people on either side of the contract.
Back Office Talent Strategy — we measure what provider attrition is actually costing you, rewrite the governance and contract terms that make knowledge an owned asset, and cut the exception load that creates the dependency in the first place.