That's probably the signal. But as one of the parent posters said, the # of folks who use such budgeting apps is quite small. For advertising, small samples are useless, so this data has to be modeled to the full US population.
For that, this very biased training set. And almost always the independent variables used for modeling are 7-10 standard demographics.
For that, this very biased training set. And almost always the independent variables used for modeling are 7-10 standard demographics.