EFA is a complicated technique. Basically items are excluded if they have low weightings on all factors, have cross weightings or have low communalities. Please see my study guide.
Reliability and validity are two main criteria for measuring the goodness of measures for a research instrument. Refer the https://chfasoa.uni.edu/reliabilityandvalidity.htm for the type of reliability and validity of the instrument.
EFA is used for data reduction and summarization. Exclude items with low loadings say below 0.4 or 0.5 based on the literature. Also, refer to the cross loadings. In a case of EFA and CFA, split the data. t is common to split data in half and to do EFA on one half and CFA on the other half.
CFA and EFA are two different approaches. CFA should be based on the theory/previous literature. You wouldn't do CFA to demonstrate that your model fits your data or to simply support your theory/model. This would be a very weak test of model fit because of its the same data as from the EFA.
Yogesh indicates that EFA is a data reduction technique. This is not necessarily true. Certainly, EFA may be used to reduce variables into a more manageable format (i.e., create latent variables called factors that reflect the underlying variables). Thus, EFA is not necessarily a data reduction technique. It may also be used to develop and refine theory.