Ad-Hoc Robust design CJS

[aptrk]

Good morning all,
I'm working on a dataset of 250 marked nighthjars and 550 surveys/sampling occasions. When the research started there was not a clear design to estimate survival, and surveys where conducted whenever it was logistically feasible. This means some surveys are separated by 50 days and other are separated by one day and are therefore grouped.
I am thinking of using some sort of robust design to estimate survival where the primary periods are somehow defined ad-hoc (maximum distance between th 1st and last survey should be one week). Two questions:

1) Does this make sense?
2) Ideally I would like to estimate something like monthly survival but given the timing of the surveys that may be challenging right?
I guess I could pick a few days a month as primary occasions but that will mean losing lots of data.

I would appreciate your feedback on these ideas, thanks

Alejandro