How to benchmark emerging AI roles – where salary data is limited

Emerging roles (like AI Engineers, like AI data scientists, prompt engineers, or VP of AI in 2026) typically have one thing in common: a lack of benchmarking data.

That leaves People and Compensation Leaders with a big headache. How do you set competitive compensation ranges when the underlying data on market competitiveness is lacking?

We spoke to the compensation experts Jörn Diekmann (Independent Total Rewards Consultant and member of rewards) and Alistair Fraser (Founder of Justly and member of rewards) about their approach to benchmarking new and niche roles, where reliable data is scarce.

Here’s their advice in a nutshell:

1. First, understand the role and gather data from all relevant sources
2. Then, analyse the data gathered and look for patterns
3. Start with existing compensation frameworks, before evolving as emerging roles mature
4. Review regularly and adjust based on feedback from the market

Let's take a closer look.

First, understand the role and gather data from all relevant sources

“When hiring for roles in emerging fields (like emerging AI jobs today) it’s essential to first ensure a clear understanding of the scope of the role,” advises Jörn Diekmann.

This is crucial for many reasons, including ensuring you identify the best-fit candidate. Specifically, when it comes to benchmarking, jumping straight into the data without first clarifying the desired scope makes it very difficult to select the most appropriate dataset and to accurately compare market data with your internal needs.

From there, both Jörn and Alistair recommend exploring multiple different compensation data providers to gain as comprehensive an understanding as possible of the market for the role, and enable meaningful comparisons.

"Over-relying on only a few isolated data points or anecdotal evidence from peers is a big risk for an emerging role with limited market data," says Jörn. “It can result in overpaying for trending skills, or underpaying and struggling to attract and retain critical talent.”

"You wouldn't accept the first price when shopping for a new car”, explains Alistair Fraser. “You'd compare models, check different dealers, maybe even get a second opinion.” 

“The same goes for compensation. When benchmarking an emerging role, relying on a single data source is like buying blind. Cross-checking multiple sources helps you spot outliers, understand the range, and make a more confident decision."

If you’re struggling to find enough data, it can be helpful to explore proxy roles – identifying the closest existing role that shares a similar scope and set of responsibilities, and use that as your baseline. For instance, an AI Ethics Officer might be close to a Compliance Manager or an AI Engineer might map to a Software Engineer.

If you do use proxy roles, be mindful of market trends and competitiveness.

In these two examples, AI is currently the hottest skill on the tech job market, so you will likely need to apply a market premium to that baseline. 

Then, analyse the data gathered and look for patterns

Once you've gathered data from multiple sources, the next step is making sense of it all. 

First, ensure you’re mapping like-for-like across all the sources you’ve compiled.

"Make sure you are comparing apples to apples across each dataset, and between the data and your internal frameworks," explains Alistair. "That means matching roles not just by title, but by underlying scope, level of seniority, and job content. Job titles vary wildly between companies, so relying on titles alone can lead to misleading conclusions."

This is especially critical when you're looking at data from different sources, as each provider may categorise or level roles differently. 

"Once you've done that groundwork, look for patterns," advises Alistair. 

“Is there a commonality in what candidates are asking for in interviews? Can you see a consistent ratio in how the role you’re hiring for compares to another similar role? If you can find the clear trends, it gives you a basis to apply pragmatically to your own framework."

Alistair’s suggestion to compare with other roles is particularly helpful to move beyond individual data points and understand broader market positioning. 

For instance, if you’re hiring for an AI Engineer role, you could look across the data sources you’ve compiled and see whether there’s a consistent relationship between the typical salary for an AI Engineer and for a Software Engineer. If you see that across the data sources an AI Engineer is typically paid 10% higher than the Software Engineer, then applying a 10% premium to your existing Software Engineer salary band might be the right approach to take.

Whilst this provides helpful guidance, Alistair warns that it’s crucial to remember that this is a directional indication of trends only (also why regular reviews are a must, as we’ll see later).

"Be clear that those figures are directional, not absolute," Alistair says. "The goal when you have a lack of market data is to be as fair and explainable as possible, not mathematically perfect.”

Start with existing compensation frameworks, before evolving as emerging roles mature

"For novel and business-critical roles like the majority of the emerging AI roles currently, creating distinct roles can help attract talent with niche and emerging skills, and signal organisational commitment in the market," explains Jörn.

However, both experts recommend staying within existing career frameworks and compensation structures until emerging roles mature – with job titles, role scopes, and career paths becoming more formalised and standardised. 

"Creating a completely new structure from scratch is rarely necessary," explains Alistair. "A more pragmatic approach, especially when the role is adjacent to an existing track, is to apply a market-driven differential to the closest existing band you have, based on the research you’ve done."

For example, if you’re hiring for your first ever AI Data Scientist this year, you might decide to stick with your existing Software Engineering job architecture, job levels, and salary bands to maintain internal consistency whilst there are uncertainties about the emerging role – but apply a market premium to account for the comparisons you’ve seen in the data, reflecting the in-demand nature of AI as an emerging skillset.

“As soon as new roles commoditise and become more mainstream, review your job architecture and compensation framework and consider formalising any new roles to ensure full alignment with external market practices” says Jörn.

Review regularly and adjust based on feedback from the market

Regular market reviews are essential for all roles, but for emerging roles the review cycle may need to be more frequent, because shifts are more likely to occur due to the uncertainty of the salary data you’re relying on.

"As with all roles, compensation should be reviewed on a cadence that makes sense for your business – typically quarterly, biannually, or annually – to ensure you remain competitive in the market you are operating in, and ensure your data source(s) are still fit for purpose" explains Alistair. 

However, Alistair cautions against overreacting to every market movement – we know the data we’re using is more volatile, which means that changing an individual’s salary to reflect every market move is also more risky. 

"Just because the market rate has moved does not automatically mean you need to follow it,” Alistair says. “If you are still able to attract and retain talent with your current pay strategy, and it is working for your business, there is no reason to change it. Market data is one input – not a mandate."

So, review the data regularly, but also keep an eye on real-world indicators that suggest your compensation positioning may be off.

"Look at how many offers are being rejected and whether you are seeing higher attrition in that role, especially when people are citing pay as a reason for leaving," advises Alistair. "With that information, you can make an informed decision on whether to adjust compensation for the role."

Overall, the key is staying flexible and data-driven, maintaining internal fairness and adjusting the approach as both the external market and your internal understanding of any new role evolves.