4 Patterns in Linkedin Search Results for Software Engineers
Perhaps one of the most fascinating things about Machine Learning is its capability to effortlessly detect patterns that one would not normally look. In the process of developing our app, which evaluates the likeliness of an engineer in being interested in a given new opportunity, we’ve stumbled across a few interesting patterns of Linkedin search results:
1 – More weight to candidates that have not changed jobs recently
While this might sound logical and obvious it’s still good to know that such filters are being applied by default. Only 9.6% of candidates in the search results had changed jobs less than 3 months ago. While 80% changed jobs at least 6 months ago.
2 – More exposure to candidates that have never been promoted.
While it sounds pretty obvious that if a candidate has been promoted he is less likely to move one could also argue that candidates that do get promoted and are happy about it simply don’t update their linkedin profile, contrary to candidates who do get promoted but are disappointed with the terms of the promotion. Regardless to either argument 86.4% of candidates in the search results had supposedly not received any promotion in their current company.
3 – Re-locations seem to be irrelevant
One of the functions our application does is to determine whether a candidate has recently relocated a given distance in order to evaluate whether he/she might be open to relocation. The distribution seems quite even if we assume that a recent relocation is anything less than 2 years from the current date.
4 – Even distribution in number of jobs (Seniority?)
Another interesting pattern, which is hard to interpret as a pure coincidence, is that the number of jobs a given candidate had is quite evenly distributed. Could this mean that Linkedin puts an effort in showing candidates of different seniority? (assuming more jobs = higher seniority, which could very well be a wrong assumption)
Our core mission at CV Timeline is to simplify and delegate the complexities of evaluating a candidate’s likeliness of interest in order to decide whether you should invest the not-so-little time required to effectively reach out to them. Get our free chrome extension.
10.000 profiles generated in Linkedin Recruiter search results for search queries containing only technology keywords and location (“i.e. Java Developer, London”) without any other filters, with an even distribution of programming languages, and across 20 different cities in 12 different european countries.