Stop deciding pay oneawkward conversationat a time.
Two engineers doing the same work on salaries eleven thousand apart. A counter-offer you agreed at 9pm. A candidate asking what level the role is and nobody able to answer.
Book a call →Pay drifts quietly, then all at once
Nobody sets out to create a mess. It happens because the first fifteen salaries were set by negotiation, the next twenty by whatever the market was doing that quarter, and the rest by matching whoever shouted loudest. Two years in you’ve got people at the same level thirty per cent apart and no way to explain it.
What it costs you →
The cost shows up in three places. Offers get slower because every one needs a debate. Good people leave because the only reliable way to get a raise is to resign. And when someone asks why they earn less than the person next to them, you have no answer that holds up.
What is job levelling?
Job levelling is the process of grouping roles into job families and defining the levels within each one, described in behaviours rather than years of service. It is the step that has to happen before salary benchmarking, because you cannot match a role to market data until you have agreed what that role actually is.
Structure before numbers
You cannot match a salary to the market until you’ve agreed what the job actually is. So levels come first, written in behaviours you can point at rather than years served.
Learning the craft
Delivers alone
Owns outcomes
Sets direction
Shapes the org
Four to six weeks
What you end up with
- Job family and levelling framework, written in behaviours not tenure
- Market benchmarking against your sector, stage and locations
- Salary bands with deliberate width and overlap
- Gap analysis showing who sits outside band, and what fixing it costs
- Pay review and promotion policy
- Manager guidance for the conversations that follow
- Offer guidance so new hires land in band from day one
What founders ask
What data do you benchmark against?
Recognised market data matched to your sector, funding stage and locations, cross-checked against live hiring data. National averages are close to useless for a venture-backed tech business.
Do we have to publish our bands?
No, and plenty don’t at first. What matters more is that managers can explain how pay is decided.
What about people already paid above band?
Usually not a crisis. Hold them flat, adjust the band if the market has genuinely moved, or treat it as a documented exception. What you can’t do is pretend it isn’t happening.
How often should bands be reviewed?
Annually for most companies, and again after any raise that changes your hiring plans. Tech pay moves quickly enough that two-year-old bands are fiction.
We’re only thirty people. Too early?
Thirty is roughly right. Early enough that fixing gaps is affordable, late enough that you know what the jobs are.