Subscription or Fee May Be Required to Access the Information on the External Link
Released in October 2025, the Job Ad Analysis Toolkit (JAAT) is an open-source collection of processing tools designed to extract standardized and structured information from online job postings. The tool was developed in collaboration between Loyola University Chicago's Quinlan School of Business and the Technical University of Munich.
JAAT is built around the Occupational Information Network (O*NET) framework, which is the federal government's primary database of occupational tasks, skills, and requirements. It uses this framework as a structure to classify and label the unstructured text of job postings. It aims to allow researchers to better understand commonalities in job postings, even when those postings differ widely from one another. Using JAAT and data from the National Labor Exchange (NLx) Research Hub, researchers extracted more than 10 billion data points from more than 155 million online job ads posted between 2015 and 2025. The resulting data is a large-scale public-use resource for workforce researchers, policymakers, and education and workforce development practitioners.
The National Labor Exchange Research Hub is a federally supported repository of job postings data that, unlike commercial vendors, makes its underlying data openly available to researchers at an affordable cost. This openness makes it a natural foundation for tools like JAAT, which are designed to show their work rather than obscure it.
The toolkit includes six core tools:
Job postings data have become one of the most widely used sources of real-time labor market information. Policymakers cite job posting counts as evidence of labor demand, while many workforce agencies use them to build career pathway tools and identify in-demand skills. Researchers often use them to track credential requirements, wage trends, and the spread of artificial intelligence across occupations, while reports on job posting data are popular topics of reporting across media outlets.
However, job postings are not typically designed for research purposes. Most are written to attract applicants and comply with organizational and legal norms. A typical job ad contains fewer than 200 words related to skills out of roughly 1,000 words total, and only about 40% include information about education level. Few include reliable salary data. The result is that two vendors analyzing job ads from different listings sites, or even vendors using different methodologies to analyze the same site, can produce different pictures of employer demand.
JAAT was developed to address these limitations by providing an open-source, transparent alternative to labor market data products. Researchers and analysts using JAAT can examine and replicate the methods used to extract data, assess accuracy, and identify sources of error.
Meisenbacher, S., Nestorov, S., & Norlander, P. (2025, October). Extracting O*NET features from the NLx corpus to build public use aggregate labor market data [Working paper]. Washington Center for Equitable Growth. https://equitablegrowth.org/working-papers/extracting-onet-features-from-the-nlx-corpus-to-build-public-use-aggregate-labor-market-data/
Quinlan School of Business, Loyola University Chicago. (2025). Job Ad Analysis Toolkit (JAAT) [Software repository]. GitHub. https://github.com/Job-Ad-Research-at-QSB-LUC/JAAT
Albert, K., Strohl, J., Daniels, P., & Norlander, P. (2026, June 17). Job postings aren't jobs. Work Shift. https://workshift.org/job-postings-arent-jobs/
The AI Library Mapping Assistant is a work in progress. Users are advised to verify the information from additional sources.
We strive for accuracy but acknowledge that errors may occur, and we encourage users to provide constructive feedback.
Have something to add or refine? Your input in this work matters greatly and we look forward to reviewing your additions
