Why the System Needs an Overhaul

Recruiters and job seekers alike rely on LinkedIn’s Skills section to match talent with opportunities. But many users report frustrating mismatches.

Simple variations like typos, abbreviations or even U.S./U.K. spelling differences can prevent a candidate from being a better match to a job. 

For example, one job seeker analysed 12 similar job postings and hit LinkedIn’s 50‑skill cap after only 6 jobs. By 12 jobs he had 75 unique skill names – most never appearing on his profile – and noted “even crazy spellings of the same thing… This isn’t helping anyone”.  Another commenter bluntly called the matching “trash”, asking “who on earth has only 50 skills after years of working?”.

LinkedIn has increased the limit to a 100 in February 2024 and in another post from March 2023 they detail how they deal with skill taxonomy and skill graphs, but it seems to me that something is not working as intended.

Common Skill-Matching Pitfalls

• Typos and misspellings. A simple typo (e.g. “Pythno” instead of “Python”) currently means no match, even though a human reader would understand. Recruiters often end up with very rare or misspelled skills in their job requirements (even “animal” as a listed skill), forcing applicants to waste a skill slot or gameplay the system.
• Singular vs. plural. LinkedIn treats “Analyst” and “Analysts” as distinct skills. A recruiter searching for “Data Analyst” may miss a candidate who (correctly) listed “Data Analysts” on their profile.  This small difference can block otherwise perfect matches.
• Abbreviations and acronyms. Many fields use shorthand (e.g. “ML” for Machine Learning, “CRM” for Customer Relationship Management).  LinkedIn’s taxonomy should include common aliases, but the system still struggles.
• US vs. UK spelling – locale issues. Words like “organize” vs. “organise” or “behavior” vs. “behaviour” can fail to match, even though they mean the same thing. LinkedIn’s skill database spans many locales (they report “374k aliases across 26 locales” ) – yet users find that English variants often act like completely different skills.
• Bundled skills. Some candidates combine skills into one phrase (e.g. “Agile Project Management” instead of listing Agile and Project Management separately). This hurts both sides: recruiters searching “Agile” might skip a profile that only shows “Agile Project Management,” and vice versa.  As one commenter complained, hiring managers “break out one term into many micro terms… or call it something silly,” forcing candidates to list even basic tools (“do I really need to list… Microsoft Word?”) .
• Skills vs Languages. Some job postings list skills such as English, French, those are under the languages section and LinkedIn should provide a way for job posters to match.
• Skills vs Tools. Should there be a tools section? Some jobs will list Microsoft Word as skill (and of course all of its variations: MS Word, Word, etc, etc.)

Side note: I had the same skill twice, once because I have added it myself and a second instance because I had completed a LinkedIn training. LinkedIn still asks me once in a while if I speak English to match me with better jobs 😁

These issues make LinkedIn’s skill-matching brittle. In practice, that means job seekers feel penalised for minor wording differences and can’t rely on LinkedIn’s keyword match to reflect their true qualifications.

LinkedIn users have sounded the alarm on forums, it does not take a long time researching to find these posts.

Toward a Better Skills System

It’s clear LinkedIn’s Skills Graph (with ~39,000 skills and hundreds of thousands of aliases, numbers from 2023) needs smarter handling of real-world variability.

• A validated, multilingual taxonomy. LinkedIn should further refine its skill vocabulary by validating entries and supporting multiple languages/locales. Every skill on a profile ought to map to a canonical term. (LinkedIn’s own engineering blog mentions they track aliases and translations for each skill – this should be used to the fullest.)
• Automatic normalization. The system should auto-correct or match common typos, handle plurals/singulars, and expand abbreviations.  For example, searching “organize” should also find profiles listing “organise”, and “Big Data” should match “big-data” or “BD” if those are known aliases. Essentially, rely on fuzzy matching and the taxonomy’s aliases to ignore trivial differences.
• Lift or raise the skill cap. LinkedIn recently doubled the skill limit from 50 to 100 , recognizing users often hit the ceiling. This was a good step, but even 100 may be too low for multi-disciplinary professionals. Ideally, allow more skills or remove hard caps to fully showcase diverse skill sets.
• Disallow or split bundled skills. The platform should discourage cramming multiple skills into one phrase. During input, LinkedIn could prompt users to split “Agile Project Management” into separate entries, or auto-split company-specific jargon. This ensures each distinct skill is searchable on its own.
• Cross-match related skills. Sometimes related but not identical skills (e.g. Scrum vs. Agile, or “CPR” vs “First Aid”) should count as partial matches. LinkedIn could mark certain skills as closely related in the taxonomy, so that having one can boost a match for the other when relevant.
• Disallow skills from other sections. More specifically for the languages but a separate section for software tools would be certainly helpful.

Implementing these features would make talent matching fairer and more accurate. By handling variations automatically and letting similar skills overlap, LinkedIn could honor more of a candidate’s true expertise and save recruiters from missing out on qualified people.

Share your experiences in the comments. Have you hit a skills limit, or lost a match due to a typo or phrasing?

Photo by Andrea Seiler on Unsplash