As far as I'm aware it's a unit-less measure, although, this does not make it arbitrary. The strength of the importance of the coefficient is measured by the p-value, so the convention is to mention the coefficient and it's relationship between the independent and dependent variables (which is which, so which way the coefficient applies) and to describe the p-value in parenthesis besides this to illustrate whether the regression was worth looking for.
NO, the strength of the importance of the coefficient is NOT the p-value, do not confuse p-values with effect sizes! The p-value is an estimate, under the assumptions of the model at hand (e.g. normality, heteroscedacity and so forth), of how improbable the observed beta value would be if the null hypothesis (i.e. beta = 0) were true. One of the most common misconceptions of p-values is to interpret them either as effect sizes or probabilities of the effect at hand. Although effect size matter for p-value, other factors such as sample size and error variance are more important. You can have the same effect size in a 20 n sample data set as in a100 n data set but still only a p
A standardized beta coefficient compares the strength of the effect of each individual independent variable to the dependent variable in the regression. Nothing more nothing less. Regards,