In general, it's very good. For example, when I got started, I had around 100 customers in my database (each representing a specific truck type). I sent them a survey and asked about their preferences, i.e. which truck type did they prefer, diesel vs. CNG etc. They responded and it took less than two weeks to send out the survey and get back responses. Then I analysed those responses, including data about truck size, average load weight etc. And I was able to predict which trucks would respond best to my ad campaigns, and to which customers! This saved us lots of money, because we didn't have to buy ads for the wrong customers. And it helped us save on fuel costs too. And the most important thing - our drivers got happier, because now they could choose the truck type that suits their driving style the best. Also, we created a lot of surveys, based on user input, for each of our clients.