内容概览
A salary band is defensible when you can say where each number came from. Published salary guides rarely give you that: they average across employers of different sizes, across cities inside one market, and across job titles that mean different things in different companies. They are a useful sanity check and a poor basis for an offer. Building a band for one role in one market is a different exercise, and it can be done from signals you can observe directly.
Why are published salary guides often wrong for your role?
Because the average of a wide range is not a number anyone is paid. A guide that reports a band for "Marketing Manager, Singapore" is summarising roles that differ in team size, budget ownership, sector and seniority — and in whether the title is inflated. Two organisations using the same title can be six figures apart for reasons the guide cannot see.
Guides are also historical by construction. They describe what was paid over a period that has already closed, in a market where the last three months are what decides whether your offer is accepted.
What signals give you a real band?
Five, in roughly descending order of usefulness.
- **Live offers in the same market and function**, including the ones that were declined and why. This is the only signal that reflects what candidates are actually accepting today.
- **What your own people are paid** for equivalent work, adjusted for tenure. If your new hire would arrive above three internal peers, you have found a problem worth solving before you make the offer, not after.
- **What candidates currently in process are paid now**, and what they say it would take to move. Treat individual figures as data points, not as the market.
- **Advertised ranges** in markets where pay transparency is required or customary. They are a floor rather than a midpoint.
- **Public filings and regulated disclosures** for senior roles in listed companies.
How do you build a band from those signals?
Start with the work, not the title. Write down what this person will own, then find three or four roles you know the pay for that own something comparable. Set the midpoint where you would expect a solid, immediately effective hire to sit, and put the range around it wide enough to pay for something specific — a scarce technical skill, a language your desk needs, direct competitor experience.
Then state what each end of the band buys. A range with no attached criteria becomes a negotiation, and the strongest negotiator gets the top of it regardless of what they bring.
How do you test a band before you advertise it?
Before the role goes to market, have someone who works that desk read it against live activity, and take seriously the answer if it comes back low. Two cheaper tests are available to anyone:
- Ask the last three candidates who declined a comparable role what it would have taken. Declines are more informative than acceptances.
- Look at how long comparable roles stay open on job boards. Roles that sit for months are usually priced below the market, not badly written.
What do you do when the band turns out to be wrong?
Decide which of the three levers you are willing to move, and say so internally before the next candidate conversation: raise the band, narrow the requirements, or accept a longer search. Those are the only three options, and refusing to choose picks the third by default.
If you raise the band, look at your existing team in the same breath. A new joiner paid above longer-serving colleagues doing the same work is a retention problem you have created and will pay for twice.
How often should you re-benchmark?
For roles you hire continuously, every quarter, because your own offer and decline data refreshes that fast. For roles you hire rarely, benchmark at the point of hiring and do not rely on a number from the last time you ran the search.
One caution on process: in several Asia-Pacific markets, asking a candidate for current salary as a condition of consideration is restricted or discouraged, and in some it is prohibited outright. Build your band from role value rather than from the candidate's history — it is better practice, and it avoids carrying someone else's pay decisions into your structure.
Where can you get a second opinion?
Ask a consultant who is placing into that function in that city this quarter. We will tell you where your band sits against live offers, and if it is under, we will tell you that before you advertise rather than after three declines.