Many years ago, I was contacted by a skills-based hiring tech company for a consulting project. I thought they wanted advisory help, but the call made their request seem a lot more complex. They said I would be writing a skills architecture for the first skills-based hiring platform that would use skills to reject and approve candidates. HR tech companies love calling themselves the first even when they are nowhere near the first.
Admittedly, I thought the project would be a little over my head. When I think of skills architecture, I think of complex maps of how skills connect inside an organization to produce the outcomes they promise. The name implied a Deloitte project to me. So, I asked them to give me a sample of the data they wanted to create. Y’all. They sent me a spreadsheet with over 500 buzzwords listed down one side and a column for a definition. That was it. That was the data they were going to use to “create” the “first” skills-based hiring platform. This was what they thought a “taxonomy of skills” would include. I just shook my head.
After a few calls to friends who know more about AI than me, I declined the project. I just didn’t think it was ethical. At worst, it could cause harm and that’s not ok with me. I couldn’t live with knowing I contributed to a system that was rejecting and approving candidates based on my bias and singular definition of a word. At best, it was another useless technology no one would buy.
Why HR Tech Companies Love Skills-Based Hiring
I’m not trying to exaggerate when I say, “cause harm.” No, I’m not talking about people dying kind of harm. I’m talking about the kind of harm where we oversimplify data used for predictions and it has broad and deep impacts. That’s why everyone gets rightfully worried about how OpenAI is trained. Bad data, bias outputs.
It’s also why I don’t think anyone should be investing heavily skills-based hiring technology: it is an oversimplification of skills taxonomy at work that will have broad and deep impacts on hiring at your company. Look, I know why HR technology companies love it. It gives them structured, simple data they can work with. The standard job post is useless. The content is not detailed enough to automate anything.
The technologists know they need some other system to categorize information about roles in order to sell you an automation value proposition. Enter skills-based hiring tech. This oversimplification means it’s not reliable in predicting the right hire. If it does work, it’s because of bias or luck. The only thing I guarantee if you use it? You likely hired someone who is similar to the person who built the survey. Not sure that’s a winning formula for a great hire.
Collecting Good Data: The Skill Recruiters Actually Need
Whether the solution is skills-based hiring or something else entirely, the answer must address the real issue that plagued us long before we tried to AI everything: hiring managers often can’t define what they want until they see a candidate they do not want to hire. No machine can fix that. AI can’t tell you that you didn’t define the criteria correctly and if you get that part wrong? Well, there’s no chance you can get the hire right from there.
That’s why recruiters need to be focused on collecting good data, not generically categorizing jobs. Every recruiter needs to know how to get managers to tell them what they are really looking for and how to spell that out in a clear way. Then they can use AI to find the right candidate.
You get better at gathering this data as a recruiter by:
- Having high quality conversations. Yes, that means recruiters have to run a hiring manager intake every time. But I don’t mean just scheduling the meeting. They have to ask great questions (my template for hiring manager intake questions here). The only tool you need for this is training and a transcription service.
- Identify high quality data sources. Encourage your team to critically consider what is good and bad data inputs if they want to generate results. Good data points might be conversations, performance reviews, and interview recaps. Generic job posts may not make this list.
- Create a skills matrix. In short, this just means that you take time to define a role’s lifecycle from the most junior qualifications to the most senior. What is the difference between the roles? What must they know how to do to be at level 2 versus level 3? I love this for a couple reasons. First, it creates clear leveling so managers can’t argue about what job title to give new hires. Two, it improves the employee experience to let folks know clearly what they need to do to perform. I talk to too many teams that want top performers, but don’t treat people like driven type-a people who like to win. This system pays for itself in retention, development, and qualified leaders.
Just don’t pay $20,000 for a skills-based hiring platform when you really need a team training on hiring managers and consulting to create a skills matrix (ahem, call me).
