LinkedIn uses machine learning to classify unique job titles entered by members in their profiles into standardized titles to make them searchable in Talent Insights.

For example, Software Engineers on LinkedIn might input their titles as a “Software Engineer,” “SWE,” “Software Eng.,” or “Software Developer.” All of these member-inputted titles are mapped to the standardized job title, “Software Engineer”. This standardization simplifies searching for users by capturing all variations of a title using a single keyword.

Talent Insights groups related standardized job titles together. When searching for a job title such as “Software Engineer”, we automatically expand the search to include all related standardized job titles, such as “Senior Software Engineer” and “Freelance Software Engineer”.

LinkedIn’s function taxonomy contains 26 "functions" which are broad groupings of job titles. Each standardized job title maps to a single job function. For example, the title "Software Engineer" is in the "Engineering" function.

Functions include: Accounting, Administrative, Arts and Design, Business Development, Community and Social Services, Consulting, Customer Success and Support, Education, Engineering, Entrepreneurship, Finance, Healthcare Services, Human Resources, Information Technology, Legal, Marketing, Media and Communication, Military and Protective Services, Operations, Product Management, Program and Project Management, Purchasing, Quality Assurance, Real Estate, Research, Sales.

LinkedIn standardizes member skills into a taxonomy of standardized and searchable skills. We frequently add new skills, technologies, aliases, and language translations to the taxonomy so that we can include them in Talent Insights searches.

Member skills data includes skills from the Skills section of member profiles (populated by members themselves) and skills from the text of member profiles, including the job title, job descriptions and profile summary (extracted by machine learning into Talent Insights).

To predict whether a member has a given skill, Talent Insights  uses a model including these two data sources and other signals from the members’ profile (e.g. work history). Talent Insights will only attribute a skill to a member if the model is more than 90% confident that a member has that skill.

Job post skills data applies similar modeling techniques on the text of LinkedIn job posts to predict the likelihood that a company is hiring for a particular skill.

Talent Insights doesn't include data from member endorsement of skills.

LinkedIn determines members’ locations by mapping the zip code/postal code, city, or country that a member provides on their LinkedIn profile into geographic categories. For example, if a member provides a postal code, we can map them to the city that contains that postal code, the market area that contains the city, and the country that contains the market area.

“Market areas” are defined using government definitions of a city region. Market areas tend to share the same workers.

City or market area data is only available when more than 50% of LinkedIn members from a given country have provided a city-level location on their profile. For example, in Ireland, the majority of members have selected their location as “Ireland” and do not specify where in Ireland they live. Therefore, Talent Insights doesn’t include city or market area data in Ireland.

LinkedIn is one of the world’s largest aggregators of jobs. We source data from both jobs posted directly on LinkedIn via LinkedIn Jobs and job posts ingested from other sources, including company websites, applicant tracking systems (ATS), job boards, aggregators, and job feeds. LinkedIn has developed advanced algorithms to identify and remove duplicate job posts from ingested sources. LinkedIn’s robust process ensures the job post data reflects the current state of the labor market. 

LinkedIn standardizes job posts by extracting the title, location, employer, employment type, description, and industry information from each post. Natural language processing and machine learning models extract skills from job post descriptions and standardize job post titles to make them searchable.

Job post data within LinkedIn Talent Insights reflects the number of current open jobs matching the search criteria.

LinkedIn ingests external data on company relationships to identify and map subsidiaries of parent companies.

A member is considered a past or present employee of a company when they add a position to their profile. 

In Talent Insights, a member’s industry is determined by the industry of the company for which they currently work. Members who aren’t mapped to a company, or map to a company without an industry, are associated with the self-selected industry on their profile. Each company on LinkedIn is mapped to one industry on Talent Insights. In most cases, we use the industry assigned (usually by an admin) to the company on its company page.

In some cases, the industry assigned to the company may be too specific and not be supported in Talent Insights. For these companies, we select the lowest-level available "parent" industry and assign it to the company within Talent Insights, so that customers can conduct meaningful analysis in the product. In these instances, the industry assigned to a company in Talent Insights may differ from the industry assigned on its company page. 

LinkedIn determines the employment type of a member based on their job title or the employment type associated with the position on their LinkedIn profile. We identify employment types based on explicit wording in LinkedIn member job titles (e.g. titles containing words like “Contractor”, “Part-time”, or “Intern”).

Reports in Talent Insights are filtered by default to include only full-time employees so that intern, student, part-time, temp, and contractor employees do not inflate attrition numbers and total employee counts.

LinkedIn sources members' spoken languages from the Accomplishments and Skills sections of profiles, as well as each profile's language settings. 

Language proficiency is determined by the proficiency selected by the member on their profile. If no proficiency is selected, the proficiency is set to Undisclosed.

Note: If the member hasn't provided a proficiency level, then it will be assigned as native/bilingual.

LinkedIn infers a member’s years of experience as the time between the start date of the earliest position on their profile and the current month. If a start date isn't included on a profile, years of experience is set to null and is dropped from relevant data sets. Talent Insights searches don’t include members whose years of experience can’t be inferred. You can find the percentage of the total search excluded for that reason under the Years of Experience facet in the left search rail. 

A member is considered a past or present student or alumni of a school when they add education to their profile.

Recent graduates are defined by default as members who have listed a degree end date on their LinkedIn profile within the past four years. You can access more specific Recent Grads data by editing the time range since graduation you are searching for or modifying your settings by clicking on your profile image in the top right corner of the page.

The Employer brand tab of the Talent Pool Report includes the following sections:

• Engaged talent - Calculated based on the engagement of the talent pool with your organization. This includes member activities (like, share, or comment), LinkedIn Page visits, followers (new followers minus unfollows), and job views. All engagement metrics include both non-employee and employee engagement with your company.
• Jobs engagement - This section displays the job views and job conversion rates over the last 12 months.
• InMail response rate - This section displays the InMail response rates by your talent pool over the last 12 months.
• Employer value propositions - This section displays the culture, values, and benefits preferred by your talent pool, sourced from our latest member surveys.

Talent Pool Report data is updated monthly.