You log into Aon Radford or Mercer, search for the niche role you need to benchmark, and there's a code sitting right there.
Relief, briefly – the data exists.
"You'll see codes for roles, and you're like, phew, the code's in there, the data exists" says Alistair Fraser.
"But then you click into it and there's actually no data sitting behind that code. The job exists in their catalogue, but they haven't built up enough of a sample size to release the data."
And in some cases, that data will never be released – niche roles can have existed for decades, but so few sectors or companies have them that they never reach the threshold.
Which is why benchmarking niche roles is so hard: the data is thinner and the peer group is harder to define, but whatever numbers you land on still have to sit consistently alongside the rest of your compensation framework.
We put the problem to Alistair Fraser, reward consultant and Founder of Justly, and Sunny Chatterjee, Total Rewards Director at Showpad, to get their perspective on how to compensate niche roles fairly.
Whereas a niche role stays niche – relevant only for a sector, or a small group of companies – with an emerging role (like many AI roles right now) the available data grows as the role becomes established. Emerging roles also tend to be linked to high demand, as new skills become popular in the talent market, adding an additional perspective.
Some of the advice crosses both, but in this article we’re focusing specifically on the niche.
Looking for how to benchmark emerging AI roles specifically? Here’s one we made earlier, with advice from Alistair and Jörn Diekmann.
When your compensation benchmarking provider doesn’t have the data you need for a niche role, the instinct is to go looking for more sources.
Both Sunny and Alistair push back on that.
“The biggest mistake I see people make is seeing there’s little data available for the role and immediately jumping to ‘we need more data’,” says Sunny. “When actually what’s needed is more understanding of the role.”
Sunny's example is a baggage handler, or an ancillary revenue specialist – if you simply take the job title and search for it in the benchmarks, you might get nothing back. But if you can break the role down into its constituent parts, it enables you to take a different route.
“Really try and get down to the nitty gritty of what problem it is that this particular role is solving,” Sunny explains. “What are the tasks, what is the scope, what level of complexity are they faced with.”
Alistair agrees, describing the process as an investigation.
"It's like being a detective, solving a crime," he says. "You need to interrogate your witness (the hiring manager), asking them questions about the role, the skillsets, and whether there are any comparable roles internally.”
"Benchmarking niche roles is like being a detective, solving a crime. You need to interrogate your witness – the hiring manager – asking them questions about the role, the skillsets, and whether there are any comparable roles internally."
Alistair Fraser
Reward Consultant at Justly
This last part – internal comparables – is the point of this exercise.
If you can identify a set of roles within your company that have similar characteristics to the niche role you’re trying to benchmark, then that gives you an internal data source to use as your first point of call.
“Doing this helps you to step back and identify that there’s actually a pretty similar role internally,” Alistair explains. “Maybe it matches 70% of role X with a slight change in skills, and it matches 30% of role Y.”
Once those comparable roles are identified, you can blend the benchmarks for each together – taking the closest matched role and “adding on or removing what you feel like is a sensible and justifiable difference for the difference in skills.”
"Maybe the niche role matches 70% of role X with a slight change in skills, and it matches 30% of role Y, and that means you can create a blended benchmark."
Alistair Fraser
Reward Consultant at Justly
For these blended roles, if there’s one role or skillset that’s in higher demand – like “a People Operations hire who’s increasingly doing Prompt Engineer work” to give Sunny’s example – Sunny recommends leaning towards the higher benchmarked role.
"If budget isn't a problem, I'd go with the Prompt Engineering benchmark," he says. “You don't want to go straight to the lower data point, or you'll have a retention problem within 12 to 18 months for a priority skill.”
"If budgets are tight, I'd still base myself on that benchmark, but discount the midpoint slightly to reflect that you’re mixing two job codes.”
So, Sunny and Alistair’s core advice: existing internal data should be your starting point.
Those internal comparables don’t always get you all the way there for every niche role – and that's when it is time to look externally for more data.
"Only then would I really go towards the market," Sunny says, "because you have a formal understanding of how the business sees the role – which skills, competencies, and responsibilities matter – so you can identify the data that’s actually relevant."
From there, because niche roles have such thin data per source, it’s about building up data from multiple sources to give a robust picture, rather than trusting one.
"One data point is evidence, it isn't truth," Sunny says. So one job advert with a salary range for a similar role is not something to rely your decisions on.
"One data point is evidence, it isn't truth. So one job advert with a salary range for a similar role is not something to rely your decisions on."
Sunny Chatterjee
Total Rewards Director at Showpad
Sunny recommends the following sources:
If, after gathering data from all of these sources, you’re still really struggling and it’s a critical role, commissioning a custom survey from a benchmarking provider for the specific role and peer group is worth the cost.
"Especially for niche roles in a niche market, you likely need to go the custom survey route," Sunny says. "When you're dealing with aircraft maintenance technicians, for instance – seek out a provider willing to run it. It's really about gathering a peer group of interested companies. In the airline industry, it removed the element of doubt, and let us negotiate with works councils in a much more open way."
Whatever sources you use, Alistair’s advice is to document them as you go.
“Document all the details – the recruiter’s name and email, the number they gave you, the notes on their conversation, the call recording if there is one,” he says, “because when you present the outcome back to a hiring manager or a CFO, think of them as the courts. It's got to be watertight, and all that evidence is needed.”
And one word of warning: “Don’t rely on URLs, always screenshot the job advert – that URL will be pulled offline in a couple of months, and you’re left with nothing to show.”
Of course, each data source comes with a different level of reliability and relevance, so it’s also important to document your interpretation of the numbers and rationale for how you’re using it to inform internal decisions.
Take industry reports, for instance – they’re not produced by benchmarking experts, so they don’t have the same level of rigour.
“When you look at a report produced by a recruitment company you’ll typically see four of five levels of seniority,” Alistair says, “it will never be a standardised framework like Ravio or other providers would use to ensure like-for-like comparisons.”
“Equally, you don’t get market percentiles either. You might get a minimum, mid, and maximum range, or just a straight average. So how do you know if that aligns with the 75th percentile, or the 50th, or the 90th?”
This context needs to be documented and used to inform the final decision.
Sunny's team keeps what he calls a "Rosetta Stone" – one file matching external market data across sources back to internal roles, with every decision, deviation, premium, and exception recorded against it.
"If you lose track of all these micro decisions, you lose the structure and it becomes chaos."
Sunny also documents his confidence level for each source in this file, alongside the numbers – his way to account for the reliability issues that Alistair points to.
"I categorise confidence levels as high, medium, or low," he says. "I'm very candid and transparent about this to the business – this is my confidence level in the number, this is why, this is the context I've got."
"With bespoke recruiting reports, they tend not to share any information on which companies were involved in the survey, or their methodology. That drops my confidence from high to medium, straight off the bat."
Alongside this, Sunny will also sense check each data source by comparing their numbers for non-niche roles with internal benchmarks, and by using differentials to model an expected range for the niche role too.
"I use pay differentials,” he says. "What do the majority of my roles cost in my most prevalent market? I look at wage inflation, cost of labour differentials across regions, and model it out for myself." If a report lands in the same range, it earns more trust. If it doesn't: "I'm likely to disregard the report entirely, if I'm honest."
All of this will give you the data and the context to make an informed decision about where this role sits in the market, and relative to your other roles – and to be able to explain that decision to your team.
"I always say that the core currency we're trading in Rewards is trust," Sunny says. "You have to earn it, and keep earning it every day."
"I always say that the core currency we're trading in Rewards is trust. You have to earn it, and keep earning it every day – and that means documenting your rationale."
Sunny Chatterjee
Total Rewards Director at Showpad
A niche role is one that’s rare across the market – think aircraft maintenance technicians, zookeepers, or clinical technicians. These roles have existed for years, but never widely enough to build a reliable presence in most salary benchmarking surveys.
Start internally – check whether any existing role overlaps closely enough to use as a base, adjusting for any difference in skills. If nothing internal fits, build evidence externally through multiple benchmarking providers, recruiters, job boards, association data, and industry reports, cross-checked against each other and with reliability accounted for. Ravio's benchmarking data is a great starting point for niche tech roles because of the company’s focus on the tech space, and because the data is pulled live from company HRIS systems rather than periodic surveys, so roles too new or rare for traditional survey coverage often still show up.
Look internally first for the closest comparable role and adjust for any difference in scope or skills – if you have an existing benchmarking provider, like Ravio, your support contact may be able to help you identify the closest match in their job catalogue. If there's genuinely nothing to compare it to in your existing provider, gather evidence from other more relevant providers, recruiters, job postings and industry reports, document where every number came from, and set a confidence level before presenting it internally.
Get specific about what the role actually involves before deciding it's unbenchmarkable – the tasks, scope, and complexity, not just the title. That detail usually reveals an internal or external comparable you can adjust from, rather than starting from zero. Lean on contacts at your benchmarking provider – users of Ravio report excellent customer support and the ability to ask for bespoke mappings or analyses.
As one input among several, yes – but treat it as already out of date, since it's a lagging indicator by the time it's live. Screenshot listings rather than saving links, since job ads get taken down within months, and cross-check against other sources before relying on them. Always compare against a reliable source like Ravio’s live, continuously updated, rigorously verified benchmarking data.
If budget allows, benchmark against the higher-value role. If budget's tight, base the number on that benchmark but discount the midpoint slightly to reflect the blend – dropping straight to the lower-value role's data point risks a retention problem within a year. With live benchmarks for both roles available side by side in Ravio, that call gets made on real, current numbers rather than a guess at how far apart the two markets actually are.
A survey commissioned specifically for a role without existing reliable benchmarking data available, gathering pay information directly from a peer group of companies employing similar roles. It’s typically run by a benchmarking provider on behalf of the group, with results shared back in aggregate.
Traditional providers such as WTW or Mercer offer custom surveys, typically at a significant fee since it's a bespoke research project. Ravio takes a different approach through its peer group campaigns – for early adopters in peer groups still being built out (by industry, location, or other shared criteria), access is free while that dataset reaches a robust sample size. If there's a specific niche role or market you need aligned, get in touch with the Ravio team to see if they have an existing, or upcoming, relevant peer group campaign.
Yes – Ravio's dataset covers niche tech-adjacent roles including firmware and embedded engineering, robotics engineering, supply quality assurance, database engineering, and financial crime roles such as AML and fraud analysis within fintech.
Your monthly dose of market insights and expert perspectives
Join Andreea Lupu and Juan Novoa for an honest conversation about benchmarking – picking peer groups that actually reflect your talent market, avoiding the comparisons that quietly mislead you, and using market data to support a philosophy rather than replace one.
Engineering talent doesn't come cheap, and the market moves fast. Join Ravio to see how leading companies are building competitive packages that actually hold up.
Sunny Chatterjee and Alistair Fraser has an honest conversation about benchmarking the roles that don't fit neatly into a survey – the ones where the data is thin, the peer group is unclear, and you still need to make a call.
