Beyond Applicant Tracking Systems and AI - How to Position your Resume
Why it's still early days for AI in job application screening. And how job applicants and recruiting AI of future can be friends.
There is a buzz about employers leveraging powerful AI to screen job applicants. In this post, I'll argue why existing solutions are not there yet. But given that AI will transform every workflow (including recruiting) in the near future, I'll suggest a couple of ways you can ‘co-exist’ with AI. I'll wrap up with resume refactoring techniques to help your application stand out in a crowded field today.
We will start with a quick 101 on how recruiters screen resume. We will also learn about Application Tracking Systems, a tool recruiters use to manage recruiting workflows.
Biases in Resume Screening
Your resume serves as a preview of your career and experience. Employers read it to check if you have the desired skills for the job they want to fill. As they read, they pattern-match to their past successful hiring decisions. Inevitably, biases creep in. At the top of the recruiting funnel, where applicant volume is highest, employers optimize for speed and quality by applying their biases (mental models or filters) to select candidates. Attributes like past employer, job titles, education, etc. are popular filters used in hiring processes. Most recruiting tools are optimized around these attributes! Though simple, these attributes are brutally reductive when it comes to decomposing a complex career story. It’s the ultimate sausage making tool.
The Anatomy of Modern Application Tracking System (ATS)
An ATS is an indispensable tool for most recruiters. They use powerful parsers to extract information from resumes, store it, and make it searchable. Modern ATS have come a long way in how they ingest candidate data from sources like LinkedIn, GitHub, personal websites, Google Scholar, etc. While most companies still use configured third-party ATS, large enterprises like Amazon use proprietary solutions tailored to their specific needs.
Despite these advances, ATS haven't changed their fundamental approach to applicant management. Today they're just more efficient at converting resumes into keywords. There's some experimentation with in-built recommendations on top of search functionality. These recommendations are statistically enhanced models based on past hiring data. They are useful, but I see them returning more false positives than is desirable.
Why AI Hasn't Taken Over Applicant Selection?
At the top of the funnel, ATS employ basic qualifying questions to nudge candidates to self-opt out / self-select. Assessment tools (online coding, psychometric tests) also help filter applicants. The remaining pool is evaluated by recruiters and hiring managers using rudimentary observable traits. Recommendation models are applied, but they are a hit and miss. Large enterprises are cautious about the fallout from propagating biases through these statistical models.
For AI to accurately and unbiasedly assess a job applicant's experience and potential, it would require an order of magnitude more observable data around the candidate's experience. It would need powerful new AI models that not only make sense of the candidate's experience but also the variables in their environment, over the course of their careers. We are not there yet. With generative AI and LLMs becoming more capable, it's only a matter of time before foundation models can help make sense of hundreds of data points to augment hiring decisions.
I look forward to that future.
This mind shift will require a reset of how we think about skills, talent, and the workforce. The good news is that companies are starting to think beyond job titles, tenures, and even education degrees. And AI will play a big role in how modern ATS parse the weight of a career journey through sophisticated semantic analysis that goes beyond "winning phrases." Humans are complex, and modern AI will move them out of "deterministic and reductive boxes."
So, how do you stand out in the eyes of AI?
1. Multi-modal Resume and Verifiable Authenticity - Future AIs will analyze hundreds of signals around your work, like YouTube talks, Git contributions, scientific papers, design portfolios, to create a composite picture of your career and advise on job fit. Research suggests that a multi-modal approach to candidate assessment can lead to more accurate and fair hiring decisions (Campion et al., 2016). You can improve your chances by having a digital web trail. Future AI models will infer your work relative to prevailing external factors. You can improve your odds by positioning your work not in a vacuum but as part of a team / industry.
2. Learnability, Adaptability, and Future Potential - Future AIs will hone in on your ability to learn and deal with ambiguity. A trail of learning - from certifications to impromptu graduate courses to job-based learning - will boost signals. Invest in them. Studies have shown that adaptability and learning agility are key predictors of job performance (Pulakos et al., 2000).
Now, let's talk about the current times!
Today’s ATS, overworked recruiters, and hiring managers rely on filters and keywords to make screening decisions. It will take an exceptionally empathetic (and observant) recruiter or hiring manager to connect the dots and hire on potential.
The good news is you can exploit the nature of ATS to your advantage. Here are a few tips:
1. Clarity of Your Resume - The recruiter will extract attributes like your job title, employer, tenure, and project details. Describe your experience clearly and concisely. Remove unnecessary descriptors - a wordy role description will be a turn-off. Someone reading your resume should easily understand your scope, contribution, and impact. Research has shown that clear and concise resume formatting can improve a candidate's chances of being selected for an interview (Sautter & Boone, 2020).
2. Pay Attention to Keywords/Filters - Recruiters pay attention to keywords. Spelling mistakes can make you invisible (e.g., "Principle Engineer"). Similarly, they won't infer you work at Microsoft if you use "MSFT" in your resume, nor will they infer tech jargon. Studies have found that keyword optimization can significantly improve a candidate's visibility in the hiring process (Kessler et al., 2015). However, don’t over exploit them.
3. Leverage Familiarity Bias - When the recruiter or hiring manager reads your resume, they're trying to make sense of your work through the biases they're familiar with. You can help your case by including public-facing information about your work, like published papers or contributions to your startup's flagship service. Research suggests that providing relevant, verifiable information can help mitigate the impact of familiarity biases (Dipboye & Jackson, 1999).
References:
Campion, M. C., Campion, E. D., & Campion, M. A. (2019). Using practice employment tests to improve recruitment and personnel selection outcomes for organizations and job seekers. Journal of Applied Psychology, 104(9), 1089.
Pulakos, E. D., Arad, S., Donovan, M. A., & Plamondon, K. E. (2000). Adaptability in the workplace: development of a taxonomy of adaptive performance. Journal of applied psychology, 85(4), 612.
Sautter, J. A., & Boone, C. A. (2020). Improving resume format to enhance applicant success. Journal of Education for Business, 95(5), 317-323.
Kessler, R., Sabnis, S., Ray, B., & Sinha, R. (2015). Using keyword optimization techniques to improve academic search relevance. International Journal of Information Management, 35(4), 448-452.
Dipboye, R. L., & Jackson, S. L. (1999). Interviewer experience and expertise effects. In R. W. Eder & M. M. Harris (Eds.), The employment interview handbook (pp. 259-278). Thousand Oaks, CA: Sage.
Bull's eye, great insights.