Suppose from a total population on 10 million people...
In scenario 1:
During week 1, 1M people visited a website. 4 weeks later, 1M people visited the same website... And 100K of them were the same people.
In scenario 2:
During week 1, 100K people visited a website. 4 weeks later, 100K people visited the same website... And 1K of them were the same people.
So in principle, the website in scenario 1 might have a total estimated visitors over weeks 1 to 4 of (say) 1.25M (high frequency of visit, 1M p/w with high level of overlap).
Whereas website scenario 2 might have around 350K (low frequency of visit 100K p/w with limited overlap.)
Is there a way to calculate an estimate of total visitors per site based on the above information? Would we actually need more information to generate some sort of decay curve i.e. 8 weeks later also?
Ideally this would take into account total population (10M) and in a more complicated version might deal with different volumes between week 1 and 4.