Day 1 retention

Hi,
I'd like to get some feedback from the team, whether our solution to the Day 1 retention optimization would produce meaningful results.

* We have a game, which consists of 100 levels.
* We found out through some analytics that the first 24 levels are crucial for the D1 retention. If the user stays happy during the first 24 levels, he's more likely to return the next day.

How do we make the user happy during the first 24 levels?
We try to guess the difficulty the user prefers for every level.
Some levels might be more fun to play in hard more and some levels might be more fun to play in easy mode for the particular player.

So, before a particular level starts, we do the following:

GameTune.SetUserAttribute("current_level", 1);

var q = GameTune.CreateQuestion(
"difficulty",
string[]{"normal", "hard", "easy"},
AnswerType.ALWAYS_NEW, (answer) =>
{

});

GameTune.AskQuestions(q);


For level 2 we would ask the same question, but before that set current_level to 2
GameTune.SetUserAttribute("current_level", 2);


So my questions:
Would this approach (asking a single question for all levels) work? The expected result is to get different difficulties for the different levels, which would make the game more interesting for the player.

Is it enough to change a single user attribute in order for the ML to learn that we are asking to get the difficulty for a different level?