How to improve model performance in logistic regression in R




I have run logistic regression in R and then calculated the predictions using prob like below:

data$prob <- mylogit$fitted.values
data$predictions <- NA
data[data$prob >= 0.70,]$predictions <- 1
data[data$prob < 0.70,]$predictions <- 0


However none of the cases where there was a churn(272) cases have been predicted correctly:

As can be seen in the above image,none of the prob are below 70% and after I checked the distribution using summary,i tried to better the model using:

How do I improve the model??
This is the current status of predictions:

> auc(sensitivity(data$predictions,data$Churned))
[1] 0.5003309
> auc(specificity(data$predictions,data$Churned))
[1] 0.001836179
> auc(accuracy(data$predictions,data$Churned))
[1] 0.8983969
> auc(roc(data$predictions,data$Churned))
[1] 0.5014233


What’s your data look like and could you please explain a bit about the problem you are trying to solve ?