Same turnover number. Different story.

Start with the decision behind the number

An HR leader asks a peer group whether their turnover is too high. Before answering, ask what decision the comparison is meant to inform. Are they deciding where to investigate departures, whether a recent change deserves closer review, or how to explain a pattern to the leadership team? Each question needs different evidence.

A benchmark can help locate a question, but it does not supply the answer to every retention problem. Even a correctly matched comparison cannot tell you why someone left, whether a particular departure was avoidable or which intervention would help. Keep the conversation close to a decision you can actually make.

This SOA guide is for preparing and facilitating an HR peer discussion. It is not a regional turnover survey, a recommended turnover target or a report of what members said at an SOA gathering. The checklist and worked example are original editorial tools.

Read the definition before borrowing the benchmark

Singapore’s Ministry of Manpower defines the average monthly resignation rate during a quarter using the average number of people resigning per month in that quarter, divided by the establishment’s average employee count. Its annual figures average the quarterly figures. An annual average of monthly rates is not the same measure as all resignations over a year divided by average headcount.

MOM’s coverage notes also matter: the Labour Market Survey covers private establishments with at least 25 employees and, since 2006, the public sector. Do not assume that a published aggregate describes a small employer, a particular professional group or a regional workforce spanning several countries.

Use the relevant table’s own notes for the series, population and any seasonal adjustment. This guide does not reproduce a current market rate: its purpose is to help you choose and interpret an appropriate comparison. For a discussion outside Singapore, check that market’s source and definitions rather than treating the Singapore measure as an Asia-wide standard.

Source context: MOM — Labour Turnover: Concepts and Definitions · MOM — Labour Turnover: Source and Coverage

Bring a comparison card, not just a percentage

Prepare the following fields before a peer discussion. You do not need to circulate confidential numbers to discuss how a measure is constructed. A blank field is useful information: it shows where a comparison is not yet ready.

Ask peers to describe their definitions in ordinary language. Similar labels in two dashboards may conceal different inclusion rules. If you cannot align the measures from information you are authorised to use, keep the comparison qualitative and say why.

A worked example: when 3% does not mean the same thing

Illustrative arithmetic, not real company data: Employer A records six voluntary resignations over a complete year and an average headcount of 200. It reports a cumulative annual resignation measure of 6 ÷ 200 × 100 = 3%. Employer B records an average of six resignations per month during a quarter, also against an average headcount of 200. It reports an average monthly rate of 6 ÷ 200 × 100 = 3% for that quarter.

Both calculations produce 3%, but the periods differ. They cannot be placed side by side as equivalent annual outcomes. Neither number, on its own, establishes the proportion of the people employed on the first day who remained at the end: people may join and leave within the period.

Do not fix the mismatch by casually relabelling a monthly rate as annual. Return to the underlying definitions and authorised records, choose a common basis and document the calculation. If the required information is unavailable, say that the rates are not comparable rather than creating a precise-looking estimate.

Ask peers about conditions, not just the winning initiative

Once the measures align, explore what else differs. A peer may describe lower turnover after introducing a new practice. That sequence is a useful lead, not proof that the practice caused the change. Ask what happened at the same time and which employees the account describes.

For a cross-border HR community, begin with country and workforce scope, then ask about the work itself. Do not assume that a shared industry or an Asia headquarters makes two organisations interchangeable. A headquarters team, a shift-based operation and a project workforce may need different questions even within the same company.

Bring the conversation back to your own evidence

CIPD’s evidence-based HR guide, dated 23 May 2025, recommends drawing on professional expertise, scientific research, organisational evidence and stakeholder perspectives. It also emphasises judging whether evidence fits the context. This is professional guidance, not a study showing that SOA’s discussion worksheet produces a particular outcome.

Apply that distinction after the peer discussion. First, check whether your own records support the pattern you thought you saw. Then ask what employee perspectives and relevant research might support or challenge a proposed explanation. Keep competing explanations visible instead of selecting only material that agrees with the most persuasive person in the room.

For example, if a peer’s experience suggests that role expectations deserve investigation, review your own appropriate evidence about that issue. Do not announce that unclear expectations caused your departures simply because the story sounded familiar. A useful next step may be a better question for an existing listening channel, not a new programme.

Source context: CIPD — Evidence-based HR: Make better decisions and step up your influence

Protect the conversation while preserving context

Check your organisation’s permissions and the host’s sharing rules before bringing material into the room. Do not circulate employee-level records, identifiable exit comments or commercially sensitive information. Removing names may not be enough when a small team, distinctive role or departure date makes someone recognisable.

A useful discussion can work with public source definitions, a clearly fictional example or a high-level question. Do not pressure peers to disclose data to participate. Avoid recording, transcription or uploads to AI tools unless the information, participants’ permissions and applicable policies allow them.

Do not turn an informal peer circle into a pooled competitor database or a negotiation of employment terms. If a formal benchmarking exercise is needed, use an approved process with appropriate specialist review. This article is a discussion guide, not privacy or competition-law advice.

Leave with one testable question

Before leaving, write a short decision note: what you can compare, what remains uncertain, which explanation deserves investigation and who will check it. Set a review point appropriate to the evidence you need. A discussion is productive when it improves the next decision, not when every participant leaves with the same solution.

When briefing colleagues afterwards, distinguish the official source, the peer’s account and your interpretation. Do not describe a handful of conversations as an industry norm or a regional finding. If the evidence does not support a conclusion, report the gap plainly.

• Our decision: what do we need to decide, and why now?
• Our comparison: which definitions and populations genuinely align?
• Our uncertainty: what information is missing or contradictory?
• Our next step: what will we investigate, who owns it and when will we review it?