What is ATS Resume Parsing?
Every job description has specific requirements: education, work experience, certification, soft and technical skills. Every resume and CV should contain the same information to be assessed against the job requirements. This is obvious. What hasn’t always been obvious is how recruiters can effectively assess how much a candidate's resume and job requirements align.
That’s what Applicant Tracking System (ATS) resume parsing aims to solve. Parsing a resume involves analyzing the content of a resume and organizing that information in a neat, structured, standardized, and simple way for recruiters. ATS resume parsing focuses on the key elements mentioned above and disseminates this information into categories. This makes the information more accessible, allowing recruiters to compare the best resumes’ information to the job requirements easily and accurately.
Types of Resume Parsing Software
If the terms LLM, natural language processing, machine learning, JSON and XML mean nothing to you, that’s okay! Resume parsing is an increasingly technical process with an increasing variety of types of resume parsing technologies. Though technically complicated, resume parsing is simple to use. We can list many different technical differentiators when it comes to the types of resume parsers but for those of us who aren’t engineers, these are the essential types:
1. Keyword Matching CV parsing
Exactly as the name suggests, this type of resume parsing involves scanning a resume for keywords such as education, qualification, candidate experience, skills, etc., and using those keywords to identify the key information. It also uses these keywords to assess how aligned the resume is to the job requirements.
Of course, today we have considerably more complicated keyword matching than those simplistic examples. Most recruiters know the pain of Boolean searching where you’re typing out an entire string of potential versions, potential misspellings and synonyms just to search for one keyword. This is where technology like resume parsing steps in and merges linguistics with statistics, algorithms and large sets of indexed data.
Specifically, some methods include phonetic algorithms that index words to match those that are written as they sound and those using dictionary spelling such as Soundex and Metaphone, string metrics that assess distances between words and phrases to understand their potential relevant similarity like Levenshtein distance and statistical computation such as the Jaccard Index.
For example, if a candidate misspells Python as Pythn and Python is on your keyword list, this a keyword matching algorithm can still identify it using Levenshtein. distance.
2. Rule-based CV parsing
Based on the examination of thousands of resumes, a company will manually program the most common themes of resume formats and constructions into the parser. This allows the resume parser to know what to look for and understand what it’s scanning in order to extract information like contact details.
To extract contact information, resume parsers may use pattern matching where a standard pattern for a phone number would be a parameter to identify but unlike what we humans would look for, a pattern for US phone numbers would look something like this \b\d{3}[-.]?\d{3}[-.]?\d{4}\b or \b(January|February|...|December)?\s?\d{4}\b for dates.
Consider further that a date on a resume can be related to different elements of a resume: the period of work at a company, completion of qualification, etc. To solve for this Rule-based resume parsing may also use keyword matching and contextual analyses to understand and categorize what that date is relevant to.
3. Large Language Models (LLMs) based CV parsing
ATS resume parsers that use LLMs are more advanced examples of resume parsing. They involve the use of technologies such as machine learning and natural language processing. LLMs like GPT-4 provide multiple advantages in resume parsing.
Advantages of Resume Parsing tools Using LLM
- Accuracy: Advanced language understanding allows for precise extraction.
- Flexibility: Can handle varied resume formats and structures.
- Contextual Understanding: Better at interpreting the context of information.
- Adaptability: Easily updated with new prompts and rules without changing underlying code.
Of course, having this immense power does have its drawbacks: that amount of power requires an equally immense amount of computational power. Similarly, the quality of the output of resume parsers using LLMs is equally reliant on the quality of the input prompts.
4. Hybrid CV parsing
Hybrid systems combine different approaches to resume parsing to understand a resume’s content through multiple parameters such as natural language processing, LLM, keyword matching and rule-based. As every technique naturally has its strengths and weaknesses, combining multiple techniques allows a much more accurate parsing while also allowing much more flexibility in how candidates can present and write their resumes.
Hybrid parsers focus on achieving this balance by ensuring adaptability and context-awareness using multiple techniques. Importantly, we need to consider why this balance matters. One of the key challenges that resume parsing faces is the potential bias towards how different languages, cultures and industry norms write and structure their resumes. One of the key perceived advantages of resume parsing is reduced bias, ensuring that the technology behind resume parsers is being designed to combat any potential bias is key.
For example, it may search for keywords to guide LLM on where to focus its analyses. And to analyze whether the use of the keyword is accurate in the resume context. Further, it could analyze whether the keyword in the resume context is relevant to the keyword in the job requirement’s context.
Standalone Resume Parsing vs. Integrated ATS Resume Parsing
Applicant tracking systems (ATS) essentially all have built-in resume parsing features. If you’re using an ATS, using the integrated parser simplifies your talent acquisition tremendously but if you’re one of the few recruiters who use company-proprietary ATS technology, integrating standalone products is a standard practice.
What Are the Benefits of ATS Resume Parsing?
1. Standardized Format
Your candidate information is presented using the same language, in the same format and separated into the same categories. Recruiters don’t need to search resumes to find information or interpret a variety of ways that individual candidates present the same information. Having all your data standardized streamlines processes and ensures that you’re not missing key information on a candidate’s resume.
2. Reduced Bias
Recruiters spend an average of about seven seconds reviewing a resume.
Clearly laid out, visually appealing, predictably formatted and well-written resumes are inevitably more likely to be reviewed accurately and positively within seven seconds. But how relevant is the skill of designing a resume in that way really? How many job requirements include being an expert resume designer? By standardizing the way that candidate data is presented, recruiters can assess candidates equally - without irrelevant impressions purely based on a resume design or by simply missing information on a resume.
3. Time-saving
The average job posting around 2010 received 120 applications. By the end of the 2010s, that’s up to 250 applications.
Trying to review 250 documents formatted differently, using different language and highlighting different categorizations of information is time-consuming and inevitably allows for human error. With recruiters and hiring managers increasingly needing to focus on strategy, increasing manual work like this just isn’t sustainable anymore. Resume parsing streamlines and ensures more accuracy in hiring processes by reducing the time spent on manual data entry and the inevitable human errors.
4. Candidate Matching and Scoring
Returning to that topic of effectively assessing how much a resume and job requirements align, we can see how having resume data neatly laid out allows recruiters to manually assess this alignment. But it also allows technology to assess this alignment for recruiters. More and more ATSs also automatically use this extracted data to provide recruiters with suggested candidate matches for a job and a score for each candidate in relation to a specific job.
5. Increased Usability of Resume Data
Job requirements are at times quite simple to assess. A candidate needs this degree with this many years of experience: simple initial qualifying criteria. But not every job has such clear requirements. Advanced ATSs can now use this resume-parsed data to go further than simply highlight and match qualifications and years of experience. Artificial Intelligence (AI) and natural language processing (among others) can also analyze more subtle aspects of a candidate and can provide summaries of experience.
The Challenges of ATS Resume Parsing
Applicant tracking system-friendly resume has increased in Google searches by 700% in the last 5 years, with applicant tracking system checker following suit at 550%. ATS resume parsing is an understood standard practice but why are these searches increasing so much? Because ATS resume parsing doesn’t always get the dissection of information all that right. The vast majority of the time it does, but imagine being the candidate who has a resume that simply isn’t an ATS-friendly and is consistently misrepresented. Candidates are aware of this potential and recruiters should be too.
1. Keyword Saturation (and upcoming tricks)
Most people will do everything they can to ensure they represent themselves as well as possible when applying for a job. Which is a good thing but sometimes enables misrepresentation of who they are and their alignment with a job. Keyword saturation is just one example of this. It’s the overfilling of resumes with keywords to increase technology-created match scores. To the extent where websites even advise candidates to add keywords in white text to the end of their resumes so that recruiters can’t see it but technology can read it.
When qualifying candidates becomes based on technology’s preset parameters and algorithms, it also allows for ways to trick it. Luckily we also have an increase in integrated, simple-to-use assessment tools like Equip that focuses on assessing specific, role-related skills to prevent candidates from faking their way through the hiring process with tricks such as keyword stuffing,
2. Formatting
Those Google searches increasing so much is a good thing. Unfortunately, some ATS resume parsing functionalities aren’t great at understanding creative, visual or unique resume designs. While the last ten years encouraged creating resumes that stood out, resume parsing more accurately understands standard, predictable formatting.
3. Languages and Context
Bullet points, short descriptions, standard language use and predictable formatting. That’s what an ATS resume parser wants. But when a resume isn’t that, errors occur. As a recruiter, it’s important to understand this and understand that different countries, multilingual CVs and different industries have different norms. As we touched on above, this is where resume parsers that are designed to achieve the balance between extracting accurate data and, understanding contextual and structural differences are key.
Resume Parsing: A Solution to Enhancing Efficiency
The crux of recruiting is aligning candidates to jobs. Every recruitment process starts with a candidate's resume and a set of job requirements. Every recruiter wants solutions to speed up, standardize, reduce bias and ensure accuracy in their process. But once again, how obvious is the answer to achieving that? It’s not. It probably won’t ever be but technology is increasingly offering solutions to assist. An ATS is an established essential tool by now and resume parsing empowers ATSs to further solve for bias, irregularity, inaccuracy and overall inefficiency. And which recruiter doesn’t need that?