Using our heads to solve your Reward challenges.
Let's just ask AI for our pay benchmarking…
We don't doubt that this conversation is being had in organisations throughout the land - after all employees can 'google it'. On the one hand organisations want to save money and perhaps don't want to buy costly surveys, but they do want the data as it fits with the drive to pay transparency and sharing pay ranges for roles.
AI can perform pay benchmarking, and it is very tempting not only for speed and cost and increasingly used to improve the accuracy, fairness, and efficiency of compensation analysis. But there are serious limitations with it, so let's take a look at the pros and cons.
So how does AI pay benchmarking work?
AI pay benchmarking uses machine learning algorithms to analyse salary data from multiple sources, including job boards, company records, industry reports, and government data. These systems provide real-time benchmarks by continuously aggregating, processing, and comparing compensation information across different roles, locations, and industries. As a result, companies could access up-to-date and objective salary recommendations that keep pace with shifting market trends as long as the data is in the public domain.
Key Benefits of AI-Driven Pay Benchmarking
But - and here is the big But -
While AI can dramatically improve the compensation benchmarking process, organisations should still maintain oversight to avoid potential data errors or privacy issues. Or garbage in, garbage out. Try asking for a market benchmark for an unusually titled role, or the opposite, something like Project Manager which can range hugely.
We know from extensive experience that relying a job title can be hugely risky and so we ask clients for detailed job descriptions and conversations with functional leads and People Partners to help us truly understand the role.
When we benchmark jobs the 'big two; are seniority and job content. Location and other factors play a part but if the size and content are not correct then the benchmark is useless.
So here's an example that is very close to the heart of many in Reward.
We asked a major AI system - how much does a Reward Manager in London earn?
Pretty specific right? Clear job title that is well understood, definitely about Reward and manager helping to understand the seniority, also giving a specific location as we know London may pay more.
And here is the answer -
“A Reward Manager in London typically earns between approximately £65,000 and £80,000 per year on average, with some variation depending on the specific employer and level of experience. Salaries reported range from about £55,000 to £100,000, with average figures around £65,000 to £77,000 annually. Higher salaries can be seen in certain sectors or senior roles, with some postings indicating up to £90,000 or more for more experienced positions in multinational companies or financial services.”
is one heck of a range - £55k to £100k, and then within that two narrower but still fairly wide ranges of £65k to £80k and £65k to £77k are given . But what is the mean or the median? If our principle is market average, how do we get that?
And we reward people know that 'Reward Manager' can mean a range of things - how can we be sure that we are comparing apples with apples?
We would be fairly confident on the role content here as Reward is a specialist term but 'manager' can cover quite a range. And that is where the human aspect comes in and indeed why the major pay data providers have a levelling system so you can confirm that, for example, Reward Manager can be matched to at least 3 different levels within published salary data providers giving very different results.
Having up-to-date salary benchmarking data is essential to ensure your employees' pay stays fair and competitive — especially given how fast-paced the talent market is today.
So whilst AI can give a good starting point, it is unlikely to provide a robust benchmark in line with your reward principles (and we have never seen a reward principle that says - let's see what the average on the internet is today)
What other options are there?
Recruitment agencies have always been a good source of up to the minute data on the last person appointed, but it is the most recent appointees which is not always reflective of everyone in role and there is an incentive for them to negotiate the maximum salary or package for their candidates. So good for a ballpark or update on trends but not for benchmarking.
What about what is published on sites like LinkedIn or Glassdoor. It's a start but we know from experience that very few ads carry salary information, it can be ranges which are not hugely helpful - many job ads list broad ranges (e.g. £50k-£90k) because they're open to hiring at different levels - that makes it hard to benchmark pay for a specific job level. Oten it is only for spot rated or junior roles that appear. This may improve with time and pay transparency. On something like Glassdoor where users input their own salaries, this means unverified and inconsistent self-reported data. There is no way to filter by company type or size which could be very important. And we need to know how many data points are in there or we could just be looking at a handful of companies and their reward policies, not yours.
Many organisations use salary survey data from pay data providers who run large-scale salary surveys by inviting companies to submit their employee reward data. That data is then aggregated and packaged into benchmarks you can purchase, often only if you also submit your own data,
They're credible and robust but slow. And since surveys are typically run once or twice a year, the data is many often months old by the time you receive it and you may already have done your pay review. That said, the credibility can really help with leadership buy-in, which can help you with budget approvals. They have an established data pool with broad participation.
But they are not perfect - they are error-prone due to manual submissions albeit generally by reward or HR professionals. It is also very time consuming to do this.
It is also hard to compare with relevant companies as most survey data is based on a broad data pool, often skewing toward large, global enterprise companies which can make the insights irrelevant to fast-moving companies like high-growth tech scale-ups. There are sector surveys but of course with smaller data sets.
There are some specific real-time salary benchmarking tools that can gather employee salary data directly from the source by integrating with company HRIS systems — automatically updating benchmarks to reflect any changes and ensuring real-time alignment with market changes. There are some definite pros -Always up-to-date at your end, fewer errors and more accuracy as minimal manual interventions, most offer filters like location, industry, headcount, and company stage — allowing you to benchmark against peers that actually match your organisation. But getting the right data set is absolutely critical so one tool is not suitable for all. But they don't have the reputation of the legacy surveys.
So what is the answer?
With unlimited funds and time, using all of these methods would be great. But organisations need to find the right solutions for them. That is something that Reward Heads are frequently asked to do as we are independent of any provider and will always find the right data source for our clients.
But even more than the data source, the expert human is still required to ask the right question, critically challenge the 'answers' provided - can you look AI in the eye and be sure that they stand by their recommendations? Would you be willing to recommend a package for a new Exec member based on an AI search?
We would love to help you identify the right package to attract, motivate, align and retain the best people for your organisation at the right cost and ensuring fairness. Please do reach out to us to help on rewardsolutions@rewardheads.co.uk.