How can I analyse which factors are affecting our engagement metrics the most in presence of multicollinearity

Let's say we are hosting marketing events and there are 4 variables of interest associated with each event - 1. language of event, 2. whether it was an in-person or online event, 3. whether hosted for a company of individuals, and 4. platform where it was hosted. I want to see how important each factor is in driving attendance or conversion rates. But there is multicollinearity between the variables and so we don't know which factor is the actual driving …

datascience engagement event events hosting language marketing metrics multicollinearity person platform say variables