What should be the target vairable in CTR maximization problem?

I have a dataset that contains some user-specific detials like gender, age-range, region etc. and also the behavioural data which contains the historical click-through-rate (last 3 months) for different ad-types shown to them. Sample of the data is shown below. It has 3 ad-types i.e. ecommerce, auto, healthcare but the actual data contains more ad-types.

I need to build a regression model using XGBRegressor that can tell which ad should be shown to a given new user in order to maximize the overall click through rate.

What predictors and target should I use in order to train such a model? Should I stack all the ctr values in a single column and add an extra column for ad-type, thus creating as many numer of rows as the number of ad-types for each user and use the new ctr column as target? Or is there a better way?