Is it possible to isolate weekly data from rolling 28-day totals if I don't have the starting "anchor"?

Hi everyone, I’m looking for some help with a data extraction problem.

I receive a weekly report for a subscription service I manage, but the system only provides Rolling 28-day totals. For example:

Report 1 (March 1st): Shows total revenue for the last 28 days.

Report 2 (March 8th): Shows total revenue for the last 28 days.

Since these two periods overlap by 21 days, I want to work out exactly what happened in that one specific new week (the 7 days between the reports).

The Mathematical Problem: I know the standard formula to extract a new week is: New Week = (Current 28-day Total - Previous 28-day Total) + Oldest Week (the one that just dropped off)

The Catch: I only started tracking this recently. My very first report was already a 28-day rolling total, so I don't know the value of the "Oldest Week" that needs to be added back in.

My Questions:

If I have 5 or 6 of these rolling reports, is there a point where I can eventually work out a real weekly number (not an average), or will every subsequent week be "artificial" because I never knew the value of that very first week?

If I just assume the four weeks in my first report were equal (Total ÷ 4) and use that to start my calculations, how many weeks/reports does it take until that "guess" is flushed out and my weekly data becomes 100% accurate?

Thanks for any insights!