Nearly two in every three (65%) employees are looking for a job today, and the number one reason is compensation. Meanwhile, investor funding slowed 13% quarter over quarter in Q1 2022, and respected top-tier investors like YCombinator and Sequoia are now warning their portfolios to cut costs and extend runway. This advice has big implications for headcount and pay, since compensation is the greatest area of company spend.
There’s also the familiar fact that women still make $0.82 to every dollar earned by men, a disparity that’s existed for decades without shifting — and is far worse for women of color and other minority groups. This social issue persists, in part, because businesses lack tools to measure progress in their own companies and relative to their peers. With such tools, they could treat pay parity as a business metric in the same way they monitor other key performance indicators.
This article dives into how some of today’s most interesting technology challenges can address what’s at the heart of these trends, greatly leveling the playing field between employers and employees — while giving businesses insights in a matter of clicks for which they needed expensive consultants (and multi-month projects) before.
Interesting technical challenges involved in compensation intelligence
Up until this point, technology has done little to change the fact that HR has functioned in much the same way for decades. Is automating spreadsheets really all that interesting or innovative, when business leaders and employees still struggle to answer basic questions? Here are a few examples:
- What’s our best headcount plan to weather this uncertain economy?
- How much should this candidate get paid?
- What change in pay does this employee merit?
- Do I pay my team fairly?
The following three technical challenges set employers and employees up to answer these questions with data.
Technical Challenge #1
Ingest & match employees to the appropriate record in compensation benchmarks.
Traditional compensation tools are inaccurate and unreliable for three reasons:
- The average data is outdated by 12-24 months.
- The data is not collected from employers, but from employees who may inflate reported pay.
- Benchmarks compare jobs with similar titles, without accounting for important nuances like company size, location, or funding stage.
For the above reasons, the first technical challenge is in ingesting and matching employees to the appropriate employee pay record in a normalized compensation benchmark. This includes applying machine learning (ML) models for determining accurate job role and level, by comparing similar jobs within similar companies. It also includes auto-collecting information from employers via their Human Resources Information systems (HRIS) and equity systems, to ensure data accuracy and real-time relevance.
The challenge of normalizing a company’s job titling, job architecture, and competency matrix (also known as leveling framework) is immense. To do so accurately requires technical teams to enhance employee records with external data about that employee, and then to use state-of-the-art, natural language processing approaches.
The business benefit to this work is that it allows companies, with a few clicks, to confidently match employees to the compensation benchmark — and get immediate insights into market competitiveness, as well as burn & dilution. The benefit for employees is similar, empowering them to answer questions of what they should be paid according to a company’s funding stage, size, and location.
Technical Challenge #2
Complete automation of compensation benchmarks.
But even compensation benchmarks that are built with employer-reported data and that compare apples to apples are fairly useless unless the data on which they’re built is up-to-date, especially in uncertain economic conditions like today. The fact that traditional compensation data sources are more than a year old presents a problem in job markets that are fast-changing.
For this reason, the second interesting technical challenge is to fully automate the compilation of a compensation benchmark. The shift to remote has meant that using outdated benchmark data can affect a company's ability to hire and retain top talent. The challenge is building a data processing pipeline that compiles new benchmark revisions on a frequent basis and illuminates employer drift.
Six months ago, where a company was versus the market is likely far different than its relative position today, as well as where the market seems to be moving in the near term. As such, the technical work largely involves workflow orchestration with the ability to support human-in-the-loop feedback. The orchestration must process data on a frequent basis, compile revisions, and illustrate changes.
Technical Challenge #3
Presenting and operationalizing insights.
But data and benchmarks are useless if users don’t know exactly what to do with the information (and when). To be valuable, compensation benchmarks must provide specific insights and highly personalized recommendations via simple dashboards, rich visualizations, and complex workflows and approvals. All create interesting front-end challenges in interpreting and presenting data with new products like pay ranges, scorecards, intelligent offers, and smart adjustments. Such tools, for instance, could flag when an offer is likely not to be accepted, where pay disparity exists in the organization, and how a company’s pay practices compare to similar companies.
Solving interesting technical challenges is one of the best parts of being an engineer. The world of compensation creates opportunities to build pipelines to compile new benchmarks, glean and present insights from those benchmarks, including diversity, equity, and inclusion trends.
If these aren’t interesting challenges, we don’t know what is.
Justin Byers is VP of Product & Engineering at OpenComp. Over the past 15+ years, he's helped create two exits worth $800M shareholder value at Blackline (as the fifth employee and VP of Engineering) and Shopzilla. He also co-founded PointHub and RapportBoost.ai. Connect with him on LinkedIn here.
This article originally appeared on Underdog.io.