This is called your Churn Rate, and if you can predict the reason behind it you can work in having people stick around. Find the predictable reasons and gather them in an excel sheet with this format. You can then translate this into data, count how many times customers left due a particular reason, and if this has happened a significant amount of times you can categorize this reason as a prediction of churn. To prevent these predictors from happening, send an email to people that haven’t logged in for more than 30 days with examples of how they can use your tool, you can actually have this automated. Make sure you see these tips to readdress a customer that has lost interest.