More Accurate Resume Parsing and Matching — Finally!

A significant challenge of modern talent acquisition is resume parsing/scanning and matching ideal candidates to open roles, with as little manual screening as possible. According to Glassdoor, the average open corporate job in the U.S. attracts 250 resumes from interested applicants — and no HR pro has the time to do all that screening personally.

Making the task even more difficult is the inaccuracy of resume parsing algorithms in applicant tracking systems. How off are they? A recent study found conventional resume parsing algorithms are typically only 60% to 70% accurate.

This means the integrity of the hiring process could be compromised because applicants who should be considered for roles may be getting wrongly disqualified due to inaccurate scanning and parsing of resume data.

The author of the study, Esranur Kaygin, CEO of parsing software company Hirize, said in an interview that U.S. businesses lose $1.5 million each year from missing out on perfect candidates and hiring people who aren’t a good fit because the candidate pool is so small.

Besides the failure of resume parsing AI algorithms to comprehend the nuanced language contained in resumes and job descriptions, other common causes of resume parsing inefficiency pinpointed by the study are overly fancy resume design and diverse file formats. (Ever have an applicant submit a blurry jpeg of a resume from their phone?)

Limitations of Traditional Resume Parsing

According to Kaygin’s report, conventional resume parsing relies primarily on optical character recognition technology and machine learning techniques like keyword-based, entity recognition and text chunking models that focus primarily on explicit terms.

While that’s great for scanning invoices, which have the same essential structural elements — no matter where they’re from — it often falls short for extracting the subtle meanings embedded in resumes and job descriptions, overlooking significant qualitative details.

“Each resume is like a fingerprint. They’re all unique — every single one,” she said.

Generational Divide

Kaygin observed that digital native Gen Z applicants tend to attack the algorithm issue by turning to generative AI to tailor their existing resume to the description of the specific job they’re applying for. In contrast, Gen X and boomer applicants, who can recall the time before cellphones existed, focus on manually revising their resumes.

She noted the rise of resume writing/editing startup companies that target older job seekers, promising them a resume that will fit the jobs they’re pursuing.

However, the shortcomings of traditional algorithms are being shored up thanks to a new generation of resume parsing technology that’s capable of 95% accuracy. It’s a potential game changer for the way HR evaluates and matches candidates — ensuring a more efficient, merit-based hiring process.

Resume Parsing Relief for HR

Cutting-edge innovations in resume parsing are transforming applicant tracking system technology with a three-pronged integration of:

1. Advanced optical character recognition, including image-to-text conversion for complex document formats and those dreaded resumes that contain logos and pictures
2. Natural language processing that extracts and interprets semantic information, enabling parsers to understand the context and subtleties of language within resumes and job descriptions, and
3. Large language models where you can, as Kaygin puts it, “implement common sense into an AI model.” For example, you can prompt the software that “PHP” stands for “hypertext preprocessor” and refers to a software developer position.

Here’s how that benefits HR:

• Improved resume screening accuracy: With more accurate parsing and matching, qualified candidates aren’t overlooked due to algorithmic limitations. “When you increase accuracy, you will have more candidates in the same resume pool,” she said.
• Faster, more informed hiring decisions: With the ability to provide a comprehensive understanding of candidate qualifications to the software, the previously mentioned integrated approach enables HR to make better hiring decisions.
• Increased fairness: Unlike open source AI, which has been accused of having distinct race and gender biases, next-generation resume parsing is trained to make the candidate evaluation process fair, promoting an equitable and inclusive hiring environment.

When partnering with a vendor, it’s important that you get the sense that the company understands exactly how HR works, Kaygin said, adding that Hirize offers educational resources for employers to let them know what options are out there.

By addressing the longstanding challenges faced by applicant tracking platforms, this more sophisticated, integrated approach to resume parsing that’s on the market empowers HR professionals to optimize their recruitment processes, ensuring that the best candidates are identified and matched to the right roles.