HR Automation Glossary: Define Key Terms (ATS, RPA, AI)

HR automation has its own language, and the stakes of misunderstanding it are real. Confuse RPA with workflow automation and you buy the wrong tool. Conflate AI with machine learning and you cannot hold a vendor accountable for what their product actually does. This glossary provides direct, precise definitions of every core term recruiting and HR professionals encounter — built to support smarter conversations with vendors, clearer requirements for implementation, and faster decisions at every stage of automating every stage-gate in your recruiting pipeline.

Each entry follows the same structure: a direct one-sentence definition, an expanded explanation, how the concept applies in talent acquisition practice, and how it relates to adjacent terms. Jump to any term using the links below.

Workflow Automation

Workflow automation is the use of technology to execute a defined sequence of tasks automatically, based on triggers and rules, without requiring human action at each step.

In HR, workflow automation is the foundational layer beneath every other technology discussed in this glossary. It is the mechanism that sends a confirmation email when a candidate submits an application, moves a candidate record to the next pipeline stage when an interviewer submits feedback, or triggers an offer letter template when a hiring manager approves a candidate. None of those actions require AI — they require a well-defined rule and a reliable trigger.

The distinction matters because vendors routinely use “AI-powered” to describe what is, at its core, rule-based workflow automation with a modern interface. Understanding the difference lets you evaluate whether you are paying for genuine intelligence or paying a premium for a conditional-logic engine.

In recruiting practice: Workflow automation handles high-volume, predictable steps — application acknowledgment, interview scheduling sequences, status update communications, and offer documentation. McKinsey Global Institute research identifies workflow automation of predictable knowledge-work tasks as the highest near-term productivity lever available to HR functions. For a detailed look at how these sequences are built in practice, see essential recruiting automation workflows.

Related terms: Trigger, Conditional Logic, Intelligent Automation, RPA.

Robotic Process Automation (RPA)

Robotic Process Automation (RPA) is software that mimics human interactions with digital interfaces — clicks, keystrokes, copy-paste actions — to automate tasks in systems that lack native integration capabilities.

RPA bots do not integrate with a system’s database or API. They interact with the system’s front-end UI exactly as a human user would, which makes them uniquely valuable for automating legacy software that has no integration layer. In HR, this commonly means copying data between an older HRIS and a modern ATS, generating standardized reports from systems that do not export cleanly, or onboarding employees into multiple platforms by executing the same account-creation steps a human would follow.

RPA does not learn or adapt. It executes a scripted sequence, and any change to the underlying UI — a button moving, a form field renaming — breaks the bot until it is reconfigured. This fragility is the tradeoff for its platform-agnostic reach.

In recruiting practice: RPA is the right tool when you need automation but cannot wait for a vendor to build an integration, or when the system you need to automate will never expose an API. Forrester research documents RPA as the primary automation approach in heavily regulated industries where core HR systems are decades old and vendor roadmaps move slowly.

How it differs from workflow automation: Workflow automation works through integrations and APIs; RPA works through the user interface. Use workflow automation when systems are modern and connected. Use RPA when they are not.

Related terms: Workflow Automation, Intelligent Automation, HRIS.

Artificial Intelligence (AI) in HR

Artificial Intelligence (AI) in HR is the application of computational systems that perform tasks — pattern recognition, prediction, language understanding, decision support — that previously required human judgment.

AI is a category, not a single technology. The AI applications most relevant to HR and recruiting are machine learning (pattern recognition and prediction from data) and natural language processing (reading and generating human language). Both are subsets of AI. When a vendor describes their tool as “AI-powered,” the productive follow-up question is: “Which AI technique, specifically, and what data does it train on?”

Harvard Business Review research on AI in talent management consistently finds that the most effective AI deployments augment human decision-making at defined judgment points rather than replacing human review entirely. This aligns with the architecture principle in the parent pillar: automate every stage-gate first, then introduce AI at decision points where quality of judgment actually changes the outcome.

In recruiting practice: AI in recruiting most commonly appears as candidate ranking (ML), resume parsing (NLP), chatbot screening (NLP), and predictive offer acceptance modeling (ML). Each of these is a distinct application with distinct accuracy profiles and distinct failure modes.

Related terms: Machine Learning, Natural Language Processing, Predictive Analytics, Candidate Scoring.

Machine Learning (ML)

Machine Learning (ML) is a subset of AI in which algorithms improve their predictions or decisions by training on historical data, without being explicitly reprogrammed for each new scenario.

ML is the engine behind most of what recruiting vendors call AI. A candidate ranking model, for example, trains on records of past hires — their resumes, assessment scores, interview ratings — and learns to score new applicants by how closely they resemble historically successful hires. The model’s accuracy depends entirely on the quality, volume, and representativeness of the training data.

This creates a critical dependency: ML in HR is only as good as the historical data it learns from. Organizations with small hiring volumes, inconsistent data entry practices, or historically biased hiring decisions will produce ML models that reflect and amplify those problems. Data quality is not a prerequisite that comes later — it is the first requirement.

In recruiting practice: ML powers candidate scoring, source-of-hire quality prediction, time-to-fill forecasting, and early attrition risk modeling. McKinsey Global Institute research on the productivity potential of AI specifically calls out ML-based talent matching as a high-value application for recruiting functions at scale.

Related terms: AI in HR, Predictive Analytics, Candidate Scoring, HRIS.

Natural Language Processing (NLP)

Natural Language Processing (NLP) is a branch of AI that enables software to read, interpret, and generate human language in text or speech form.

NLP is the technology behind resume parsing, chatbot candidate screening, inclusive language job description analysis, and sentiment analysis of candidate survey responses. Any time a recruiting tool reads unstructured text and takes a structured action based on its meaning, NLP is the mechanism at work.

NLP accuracy varies significantly based on the training corpus and the specificity of the language domain. Recruiting-specific NLP models trained on industry job titles and skills taxonomies outperform general-purpose language models on tasks like skills extraction and role matching. When evaluating NLP-powered tools, the relevant question is whether the model was trained on recruiting-domain text or general web text.

In recruiting practice: NLP enables high-volume screening without high-volume human reading time. It also enables more consistent evaluation — the same criteria applied to every resume, rather than criteria that shift based on reviewer fatigue. For a deeper look at AI-specific terminology that extends beyond this glossary, see AI-specific terminology for recruiting professionals.

Related terms: AI in HR, Machine Learning, Resume Parsing.

Applicant Tracking System (ATS)

An Applicant Tracking System (ATS) is software that manages the end-to-end process of a single job requisition — collecting applications, tracking candidates through pipeline stages, and documenting hiring decisions.

The ATS is a transactional tool. Its core function is answering the question: “Where is this candidate in this specific hiring process right now?” It captures application data, records interview feedback, enforces pipeline stages, and produces compliance documentation. When the candidate is hired or rejected, the ATS transaction is complete.

What an ATS does not do: manage ongoing relationships with candidates who were not hired in this cycle, nurture passive talent, or maintain engagement across multiple requisitions over time. That is the CRM’s domain. Many recruiting firms conflate the two, using their ATS as a candidate database and wondering why their talent pipeline feels empty every time a new requisition opens.

In recruiting practice: SHRM research on talent acquisition technology identifies ATS adoption as near-universal among mid-market and enterprise employers, but finds significant variation in how deeply features are actually used — most teams use less than 40% of their ATS functionality. For a detailed examination of how automation platforms extend ATS capabilities, see how Keap extends beyond a traditional ATS.

Related terms: CRM, HRIS, Workflow Automation, Resume Parsing.

Candidate Relationship Management (CRM)

A Candidate Relationship Management (CRM) system manages long-term, ongoing relationships with both active applicants and passive talent across all interactions over time — not just within a single requisition.

Where the ATS tracks a candidate through a single hiring process, the CRM maintains a persistent, longitudinal record: every email opened, every event attended, every referral submitted, every role applied for and declined, across months or years. The CRM answers the question: “Who do we know, what is our relationship with them, and are they ready to engage now?”

For recruiting firms building proprietary talent pipelines — where the database is the competitive moat — the CRM is the strategic asset. The ATS is the operational tool. Firms that run all candidate management inside an ATS are functionally starting from zero with every candidate who does not immediately convert. For a practical look at CRM-style candidate management in an automation platform, see candidate management and CRM automation.

In recruiting practice: A talent CRM with automated nurture sequences keeps warm candidates engaged between open roles. APQC benchmarking research on recruiting efficiency identifies warm pipeline depth as the primary predictor of time-to-fill performance — firms with active CRM nurture programs fill roles 30–40% faster than firms sourcing cold.

Related terms: ATS, HRIS, Workflow Automation, Candidate Scoring.

Human Resource Information System (HRIS)

An HRIS (Human Resource Information System) is the central database of record for post-hire employee data — personal information, compensation, benefits, performance history, and employment status across the employee lifecycle.

The HRIS is where the candidate becomes an employee in the system of record. It does not manage pre-hire pipeline activity (ATS) or ongoing candidate relationships (CRM) — it manages the employment relationship from offer acceptance through offboarding. Payroll systems, benefits administration platforms, and performance management tools either live inside the HRIS or integrate with it.

The ATS-to-HRIS handoff is the highest-risk data transfer point in most HR technology stacks. Manual transcription between these two systems is where costly errors occur — a wrong digit in a compensation field, a misspelled name that creates a duplicate record, a start date entered in the wrong format that delays system access on day one. Parseur’s research on manual data entry costs documents error rates in manually maintained records that create compounding downstream problems in payroll and compliance.

In recruiting practice: Automating the ATS-to-HRIS data transfer — through direct integration or structured file import — eliminates the error-prone manual step and creates a complete, accurate employee record from the moment an offer is accepted. See Keap HR integrations that reduce manual data errors for applied examples.

Related terms: ATS, CRM, RPA, Workflow Automation.

Predictive Analytics

Predictive analytics applies statistical models and machine learning to historical HR data to forecast future outcomes — time-to-fill, offer acceptance probability, quality-of-hire, and early attrition risk.

Predictive analytics moves HR decision-making from reactive to proactive. Instead of analyzing why last quarter’s time-to-fill was 47 days after the fact, predictive models surface leading indicators in real time: which open roles are trending toward 60-day fills, which candidates in the pipeline have a high probability of declining an offer, which new hires match profiles associated with 90-day attrition.

The accuracy of any predictive model is a direct function of data quality and volume. Gartner research on HR analytics identifies poor data quality as the primary barrier to successful predictive analytics adoption in HR functions — ahead of technology, budget, and skills gaps. This is why data hygiene in your ATS and HRIS is not preparatory work for analytics; it is the prerequisite that determines whether analytics produces value or noise.

In recruiting practice: Predictive analytics is most valuable for high-volume recruiting functions where even small improvements in prediction accuracy — better source quality selection, earlier identification of offer risk — compound across hundreds of hires per year. For recruiting data management terminology that supports analytics readiness, see recruiting data management and ATS terminology.

Related terms: Machine Learning, Candidate Scoring, HRIS, ATS.

Intelligent Automation

Intelligent automation combines rule-based workflow automation with AI capabilities — typically ML and NLP — so that a system can handle unstructured inputs, make judgment calls based on learned patterns, and adapt its behavior over time.

Basic workflow automation executes a fixed sequence when a trigger fires. Intelligent automation can evaluate variable, unstructured inputs — a candidate’s reply email, a resume in a non-standard format, an interviewer’s free-text feedback — and choose a processing path dynamically based on what the content means, not just what format it takes.

The practical implication: intelligent automation requires both the automation infrastructure (triggers, workflows, integrations) and the data infrastructure (clean, consistent historical records) to function reliably. Teams that jump to intelligent automation before mastering basic workflow automation introduce complexity without the foundation to support it. Forrester research on automation maturity models consistently identifies foundational workflow automation as a prerequisite stage that cannot be skipped.

In recruiting practice: Intelligent automation handles edge cases that break rule-based systems — candidate replies that do not match expected response patterns, resumes that use non-standard section headings, interview feedback that requires sentiment interpretation before routing. It is the right tool for exception handling, not core pipeline management.

Related terms: Workflow Automation, RPA, AI in HR, Machine Learning, NLP.

Resume Parsing

Resume parsing is the automated extraction of structured candidate data — name, contact information, work history, education, skills — from unstructured resume documents, using NLP and pattern recognition to populate database fields without manual data entry.

Resume parsing eliminates one of the highest-volume manual data entry tasks in recruiting: re-keying candidate information from submitted documents into an ATS or CRM. Parseur’s research on manual data entry documents that the average knowledge worker spends significant time on repetitive data entry tasks that produce error-prone records — a problem that compounds in high-volume recruiting environments where dozens or hundreds of resumes arrive per week.

Parser accuracy is not uniform. Standard chronological resumes in clean PDF format parse with high accuracy. Non-standard layouts, graphic-heavy designs, multi-column formats, and resumes in languages other than the parser’s training language produce lower accuracy. Data validation rules — flagging records where key fields like phone number or most recent employer are missing after parsing — are a necessary complement to any parsing implementation.

In recruiting practice: For a 30-50 resume-per-week recruiting operation, resume parsing reclaims 10-15 hours of manual data entry per week — time that returns to candidate engagement and client development. See automating job application intake with forms for a practical implementation approach that combines form-based intake with automated record creation.

Related terms: NLP, ATS, CRM, Workflow Automation.

Trigger

A trigger is the specific event or condition that initiates an automated workflow sequence.

Every automation starts with a trigger. In HR and recruiting, common triggers include: a candidate submitting an application form, a hiring manager updating a pipeline stage in the ATS, a calendar event being created, a specific date being reached (such as a new hire’s first day), a tag being applied to a contact record, or a lead score crossing a threshold. Triggers define the “if this” in every “if this, then that” automation rule.

Trigger precision is one of the highest-leverage design decisions in automation. Triggers that are too broad fire workflows on unintended records. Triggers that are too narrow miss records that should be included. Both produce operational problems — duplicate communications, missed follow-ups, incorrect stage assignments — that erode candidate experience and recruiter trust in the automation system.

In recruiting practice: The most common trigger design error is using record creation as the trigger when a more specific condition — a tag applied, a form field completed, a stage reached — is the appropriate event. Testing trigger logic against edge-case records before deployment prevents the most common automation failures.

Related terms: Workflow Automation, Conditional Logic, Intelligent Automation.

Conditional Logic

Conditional logic is the use of if/then/else branching within an automated workflow to route records or execute different actions based on specific field values, tags, scores, or other data attributes.

Conditional logic is what transforms a single linear automation into a dynamic system that responds to individual candidate characteristics. A simple example: if a candidate’s application indicates more than five years of experience AND the role requires senior-level qualification, route them to the senior track interview sequence; else route them to the standard track. The logic evaluates data at the point of execution and chooses a path accordingly.

Sophisticated conditional logic is what allows a single automation system to handle a diverse candidate population without manual sorting. It is also where automation complexity accumulates most quickly — nested conditions become difficult to audit, debug, and update as hiring criteria change. Documentation of conditional logic trees is an operational necessity, not an optional best practice.

In recruiting practice: For applied conditional logic workflows in talent acquisition, see the satellite on Keap automation conditional logic workflows, which covers multi-branch recruiting sequences in detail.

Related terms: Workflow Automation, Trigger, Intelligent Automation.

Candidate Scoring

Candidate scoring is the automated assignment of a numerical or categorical score to a candidate record based on defined criteria — qualifications, engagement signals, assessment results, or ML-derived match predictions — to prioritize recruiter attention across a pipeline.

Candidate scoring can be rule-based (a candidate receives points for each required qualification present in their record) or ML-based (a model trained on historical hires predicts a likelihood score for success in this role). Rule-based scoring is transparent, auditable, and adjustable; ML-based scoring is more accurate at scale but requires quality training data and ongoing monitoring for bias or drift.

Scoring is a prioritization tool, not a hiring decision tool. Its value is in helping recruiters allocate attention efficiently — surfacing the top 20% of a 200-application pool for immediate review, rather than requiring sequential manual review of all 200. Final hiring decisions require human judgment applied to the full candidate picture.

In recruiting practice: SHRM guidance on AI in hiring recommends that any candidate scoring system be validated for adverse impact before deployment and reviewed periodically as the hiring population changes. Scoring criteria that reflect historical bias in hiring data will encode that bias into an automated ranking — which accelerates a problem rather than solving it.

Related terms: ML, Predictive Analytics, ATS, NLP.

Related Terms at a Glance

What These Terms Mean for Your Automation Strategy

Terminology is not academic. Every term in this glossary maps to a decision: which tool to buy, which capability to prioritize, which vendor claim to validate, and which implementation sequence to follow. The core strategic insight from this vocabulary: automation first, AI second. Build reliable workflow automation across your recruiting pipeline before adding ML-based scoring or NLP-based parsing. The data that your workflows generate and clean is the training data that makes AI work.

The parent pillar on recruiting pipeline automation applies these concepts to a full talent acquisition architecture. For implementation starting points, the guide to automating job application intake with forms covers the first workflow most recruiting teams should build — application intake and acknowledgment — and the guide to Keap HR integrations that reduce manual data errors addresses the ATS-to-HRIS handoff that is the highest-risk data transfer in most recruiting tech stacks.