1. Introduction and Scope of ATS Implementation

Applicant Tracking Systems (ATS) are specialized enterprise software solutions used by human resources departments to administer, structure, and process the high volume of job applications. The core functionalities of these platforms include centralized storage of applications (CVs/resumes, cover letters, and profile data), tracking candidates’ status throughout the various stages of the recruitment process, and automating routine processes such as initial screening, interview scheduling, and sending system notifications.

ATS adoption is widespread in modern corporate practice. More than 98% of companies on the Fortune 500 list rely on applicant tracking systems. The technology is also widely integrated into the small and medium-sized business sector due to the emergence of affordable and scalable cloud-based solutions. For example, the software provider ADP has more than 1 million clients worldwide, while a number of other leading systems collectively serve more than 130,000 companies.

The market share and popularity of individual platforms vary depending on the segment and scale of the organization:

Platform / Segment	Usage / Market Share
Workday	Leader in adoption among Fortune 500 companies; 15.9% share outside the Fortune 500
SuccessFactors	Second most widely used system within the Fortune 500
Greenhouse	Used by nearly 20% of companies outside the Fortune 500
Lever	Used by 16.6% of companies outside the Fortune 500
ADP	Serves more than 1,000,000 corporate clients globally
Other popular platforms (iCIMS, BambooHR, Workable, JazzHR, Bullhorn)	More than 130,000 corporate users combined

Given the universal application of these platforms, an in-depth understanding of ATS architecture and operational mechanisms is a critical strategic requirement both for HR departments seeking to avoid losing qualified talent and for candidates aiming for successful professional careers.

2. Mechanics of Algorithmic Filtering and Data Processing

The technical and algorithmic processing of a given application, from the moment it is submitted until it appears in the recruiter’s interface, proceeds through the following sequential steps:

1. Submission of the application and completion of knockout questions (“knockout questions”): When submitting a CV/resume, the system requires answers to screening questions (via checkboxes or short text fields) concerning willingness to relocate, years of experience, or salary expectations. An answer that does not meet the predefined criteria results in the application being automatically marked (“flagged”), signaling to the recruiter not to review the profile.
2. Scanning and syntactic analysis (Parsing): All candidate documents that pass the initial screening are sent to the ATS database, where the algorithm performs syntactic analysis (parsing). The system converts the unstructured text format of the CV/resume into structured and indexed database fields (such as positions held, chronological work experience, employment dates, education, and skills).
3. Storage and indexing in the database: The extracted information, together with the original file and accompanying documents, is stored in the candidate’s profile, allowing the candidate’s status to be tracked throughout the recruitment process.
4. Recruiter query formulation: The human evaluator uses the system’s internal search engine (which operates on principles similar to search-engine queries), entering specific parameters such as keywords, previous job titles, geographic location, and technical skills.
5. Algorithmic filtering and narrowing of results: The algorithm filters the databases according to the specified criteria. On a platform such as Lever, a query for candidates with “SEO” and “writing” skills initially retrieves 878 profiles. Adding a criterion for a specific location (e.g., Seattle) reduces the results to 3 candidates, while introducing a condition for specific experience (e.g., between 2 and 5 years) narrows the results to 1 specific profile.
6. Visualization of the final shortlist (Shortlist): The identified profiles appear on the recruiter’s dashboard for subsequent human review and interview scheduling.

3. Algorithmic Screening vs. Human Evaluation: Filters, Criteria, and AI Scoring

The approach recruiters use when searching ATS databases relies on specific parameters. Data from a study involving 380 recruiters identifies the following primary search filters:

• Skills: 76% of recruiters filter candidates based on specifically defined skills.
• Job Title: 55% search using an exact match for the position title.
• Credentials: 51% use professional qualifications as a primary filter.

Deterministic Keyword Search (The Lever Example)

The traditional filtering model used in platforms such as Lever relies on deterministic search using Boolean logic and filters. Under this method, the recruiter manually defines a combination of criteria (skills, location, employment history). The algorithm strictly matches the entered character strings against the indexed fields in the database. This approach requires the recruiter to formulate the search precisely and the candidate to include the relevant keywords accurately in the document.

Algorithmic and AI-Based Tiered Scoring (The iCIMS Example)

Modern systems are increasingly incorporating probabilistic scoring through artificial intelligence (AI). Platforms such as iCIMS integrate AI functionality for automated scoring and ranking of candidates based on their overall profile.

The mechanics of this process include:

• Grouping by tiers: Candidates are automatically divided into separate categories (tiers) based on similar numerical evaluation scores.
• Dynamic scoring for a specific position: The algorithm calculates the match individually for each specific job position, comparing the candidate’s qualifications with the requirements of the job posting.
• Review prioritization: Recruiters receive a ready-made, automatically ranked list of the most suitable profiles, reducing the need for multiple manual, sequential searches.

4. Analysis of Obstacles, De-Personalization, and Ethical Challenges

Algorithmic Barriers and the Technical Mechanism of Parsing Errors

The unsuccessful parsing of CVs/resumes by ATS is caused by the architecture of parsing modules, which process text as a sequential linear stream. All formatting elements that depart from this linear model—such as tables, text boxes, graphic images, multi-column layouts, and content placed in headers and footers—disrupt the reading sequence. The algorithm may skip, rearrange, or completely omit the data contained within them. As a result, critical information (e.g., contact details or employment dates) may not be indexed in the database, making the application “invisible” in searches and leading to its automatic rejection.

De-Personalization, Hard Filters, and the Risk of “Manufacturing False Negative Results”

The use of knockout questions creates hard threshold barriers through which candidates are disqualified without human intervention. At the same time, because more than 50% of recruiters filter by exact job title, there is a significant risk of disqualifying suitable professionals due to terminology mismatches. A candidate with identical experience whose previous position was “Social Media Specialist” may be filtered out and rejected by the system if the search is configured for “Social Media Strategist”. This creates a risk of “manufacturing false negative results” (false negatives), whereby organizations lose valuable human capital due to terminology mismatches.

Risk of Excessive Optimization (Over-Tailoring)

The initial match rate for unoptimized CVs/resumes typically ranges between 30% and 40%. In an effort to increase this index, candidates often resort to excessive keyword saturation (keyword stuffing) in order to achieve a 100% match. This excessive optimization damages the application. When the CV/resume passes the ATS filter and reaches the human evaluator, the unnatural text, overloaded with repetitions, loses its authenticity and creates a poor impression during the final human evaluation.

5. Recommendations for CV/Resume Optimization (ATS-Friendly Compatibility)

To ensure that the document is correctly parsed by the parser and remains visible in the recruiter’s database, follow this specialized checklist:

• Achieve a balanced match rate (~75%): Optimize the content against the wording of the job posting until reaching a match of approximately 75%. Avoid deliberately targeting 100% in order to preserve the authenticity of the text for the human evaluator.
• Refine job titles: Adapt the title of your previous position to match the terminology used in the job posting (e.g., from “Social Media Specialist” to “Social Media Strategist”), but only when the job responsibilities are identical. Strictly avoid introducing false information about your professional experience.
• Remove non-linear formatting elements: Remove all tables, text boxes, graphics, images, complex columns, headers, and footers from the document to ensure linear reading of the character sequence by the parser.
• Use standard typography: Use universally supported, standard fonts with a normal text size that can be read by all operating systems.
• Standardize margins: Format the page using equal-size margins on all sides.
• Standardize dates and sections: Structure the document with clearly differentiated, standard section headings (Summary, Work Experience, Education, Skills) and use a clear, consistent chronological format for employment dates (e.g., Month/Year – Month/Year) to ensure accurate calculation of total work experience by the algorithm.