[{"data":1,"prerenderedAt":181},["ShallowReactive",2],{"insight-\u002Finsights\u002Fwhat-our-first-live-benchmarks-tell-us-about-the-future-of-salary-data":3},{"id":4,"title":5,"author":6,"body":7,"cta":167,"date":172,"description":173,"extension":174,"meta":175,"navigation":176,"path":177,"seo":178,"stem":179,"__hash__":180},"insights\u002Finsights\u002Fwhat-our-first-live-benchmarks-tell-us-about-the-future-of-salary-data.md","What our first live benchmarks tell us about the future of salary data","James Morris",{"type":8,"value":9,"toc":158},"minimark",[10,14,17,20,23,28,31,34,37,40,43,46,49,53,56,59,62,65,68,71,74,77,80,83,86,90,93,96,99,102,105,108,112,115,118,121,124,127,130,133,136,140,143,146,149,152,155],[11,12,13],"p",{},"Two years ago, PayPulse began with a relatively simple question: why should employers rely on annual salary surveys when the technology exists to provide a much more current view of the market?",[11,15,16],{},"Getting from that idea to live salary benchmarking has involved an enormous amount of work. We have had to build the platform, develop a consistent levelling framework, help organisations structure their data and create technology capable of comparing roles across different businesses and industries.",[11,18,19],{},"Now that the first live benchmarking output is available in PayPulse, I have been able to see whether those individual pieces work together in practice.",[11,21,22],{},"My first impression is that they do.",[24,25,27],"h2",{"id":26},"clear-progression-across-the-levels","Clear progression across the levels",[11,29,30],{},"The first thing I looked for was whether the data showed a logical progression in pay as roles moved through the PayPulse levels.",[11,32,33],{},"Across the majority of the linear levels, the output shows clear and consistent salary steps. As the scope, complexity and accountability of a role increase, the market data generally moves with them.",[11,35,36],{},"That may sound obvious, but it is an important result.",[11,38,39],{},"A benchmarking platform can process a large volume of information and still produce something that is not genuinely useful. If roles have been classified inconsistently, different types of work become mixed together, levels overlap unnecessarily and the user is left trying to interpret noise.",[11,41,42],{},"The early PayPulse output is showing coherent progression. It is beginning to give us a structured picture of how the market values increasing responsibility across different job families.",[11,44,45],{},"This remains an early view. Several large organisations are still completing and refining their data, and we are reviewing every metric carefully. The depth and quality of the benchmarks will continue to improve as more information enters the platform.",[11,47,48],{},"Even with that caveat, seeing those salary steps emerge so clearly is extremely encouraging.",[24,50,52],{"id":51},"job-evaluation-has-never-been-an-exact-science","Job evaluation has never been an exact science",[11,54,55],{},"Before starting PayPulse, I worked as a Pay and Reward specialist and job evaluation specialist, including on pay structure design and equal pay projects.",[11,57,58],{},"One thing that experience taught me is that job evaluation is certainly not an exact science.",[11,60,61],{},"Even when a methodology assigns numbers and points to a role, there is still judgement involved in interpreting the job, understanding its context and deciding how much weight to give different factors. The final score can create an impression of scientific precision that the underlying process does not always justify.",[11,63,64],{},"Two experienced evaluators can review the same role and reach different conclusions. A panel can spend considerable time debating the meaning of a single sentence in a job description. The use of numbers does not remove subjectivity. It can sometimes disguise it.",[11,66,67],{},"That experience strongly influenced how we approached levelling in PayPulse.",[11,69,70],{},"We considered whether to adopt or recreate one of the established job evaluation structures designed primarily for human evaluators. However, those methodologies were generally created for a world of evaluation panels, lengthy questionnaires, manuals and workshops.",[11,72,73],{},"Simply digitising an existing process would not take advantage of what modern AI can do.",[11,75,76],{},"Instead, we decided to develop a levelling framework that could be understood and applied by both AI agents and people. It needed to be structured enough for technology to reason against consistently, but clear enough for a human reviewer to understand, challenge and amend.",[11,78,79],{},"The framework considers the substance of the role, including knowledge, complexity, decision making, scope and accountability. The technology can assess the evidence within a job description, suggest the appropriate family and level, explain its reasoning and identify when important information is missing.",[11,81,82],{},"The human remains in control, but much of the initial interpretation and structuring can be completed by the system.",[11,84,85],{},"Judging by the consistency appearing in the first live benchmarking output, that decision looks to be paying off.",[24,87,89],{"id":88},"a-remarkable-first-performance","A remarkable first performance",[11,91,92],{},"What has impressed me most is how well the system has performed on its first substantial set of live customer data.",[11,94,95],{},"The job descriptions came from a variety of organisations and industries. They were not written to fit the PayPulse framework, and there was no shared template or vocabulary connecting them.",[11,97,98],{},"Despite that variety, the AI matching and levelling system has sorted roles into job families and levels with enough consistency for meaningful salary patterns to emerge.",[11,100,101],{},"That does not mean every role will always be matched perfectly, nor should an AI system make final employment or pay decisions without appropriate human oversight. Ambiguous or incomplete job descriptions will still require review, and customers remain in control of their data and classifications.",[11,103,104],{},"However, the first-time performance is significant. The system has taken a highly varied collection of job descriptions and created a coherent underlying structure without requiring every organisation to rewrite its roles first.",[11,106,107],{},"This is where agentic technology becomes particularly powerful. Rather than producing a quick answer from a title or keyword, the system can follow a structured process, examine the available evidence, compare possible outcomes and explain the basis of its recommendation. It can also recognise when the evidence is not strong enough for a confident result.",[24,109,111],{"id":110},"the-opportunity-goes-far-beyond-salary-surveys","The opportunity goes far beyond salary surveys",[11,113,114],{},"Seeing the output has also made me think again about the potential applications beyond salary benchmarking.",[11,116,117],{},"When I worked on pay structures and equal pay reviews, a significant amount of each project involved collecting job information, interpreting inconsistent job descriptions, matching comparable roles and establishing a common structure before the main analysis could begin.",[11,119,120],{},"The same challenge appears during mergers and acquisitions. Two organisations may use entirely different titles, job families and grading structures for roles that are broadly comparable. Creating a shared view can be a lengthy and highly manual exercise.",[11,122,123],{},"Could technology like this dramatically reduce the complexity and duration of those projects?",[11,125,126],{},"I believe it could.",[11,128,129],{},"It would not remove the need for experienced reward professionals, legal input or proper human review. What it could do is complete much of the initial sorting, identify likely comparisons, highlight inconsistencies and direct specialists towards the areas that genuinely require judgement.",[11,131,132],{},"For equal pay work, that could mean reaching a structured first view of roles more quickly and focusing expert attention on potential areas of concern. During a merger or acquisition, it could help organisations understand how two workforces align before beginning the detailed work of harmonising grades and pay structures.",[11,134,135],{},"The opportunity is therefore much larger than producing a better salary survey. A consistent, AI-assisted understanding of work could become a foundation for reward design, workforce planning, organisational integration and people analytics.",[24,137,139],{"id":138},"the-beginning-not-the-finished-product","The beginning, not the finished product",[11,141,142],{},"There is still plenty of work ahead.",[11,144,145],{},"More organisations are finalising their data, the benchmarks will become stronger as participation grows, and we are already exploring better ways to represent the information within PayPulse. We will continue reviewing the output, testing the levelling system and improving how customers interact with the results.",[11,147,148],{},"But the first live output has given me real confidence.",[11,150,151],{},"The salary steps are appearing where we would expect them. The framework is creating consistency across highly varied job descriptions. The technology is showing that it can apply that framework at a scale that would previously have required an extraordinary amount of manual work.",[11,153,154],{},"Two years ago, this was an idea. Today, it is working inside the platform.",[11,156,157],{},"That is an exciting point to have reached, and an even more exciting foundation for what comes next.",{"title":159,"searchDepth":160,"depth":160,"links":161},"",2,[162,163,164,165,166],{"id":26,"depth":160,"text":27},{"id":51,"depth":160,"text":52},{"id":88,"depth":160,"text":89},{"id":110,"depth":160,"text":111},{"id":138,"depth":160,"text":139},{"eyebrow":168,"title":169,"emphasis":170,"sub":171},"Live, not last year","See the first live benchmarks for yourself","for yourself","Book a demo and watch your roles line up against the market, level by level.","2026-09-01","First impressions of PayPulse's live salary benchmark output, our AI matching and levelling system, and what agentic job evaluation could mean for reward work.","md",{},true,"\u002Finsights\u002Fwhat-our-first-live-benchmarks-tell-us-about-the-future-of-salary-data",{"title":5,"description":173},"insights\u002Fwhat-our-first-live-benchmarks-tell-us-about-the-future-of-salary-data","gGWe1-sI6LOgwYM1hfzsTuijjlW85B4P7QVYPXckKS0",1788597647393]