The Estimated Annual Usage (EAU) is calculated by taking the most recent 30 days of issues to kits, and multiplying that by 12 (months).  WHen the user first begins using the EAU, the factor shoudl be 100%, meaning that the full 30 days is used to extend to an annual quantity.

If the month following (after another 30 days of experience), running the EAU again will weight the second run against the first run.  So the factor for the second run should be 50% (or the equivalent of a 60 day average).

In the third month, (after a total of 90 days of data), the factor would be 33%, thus averaging out all three months and extending that average for a year's period.

If the user wished to maintain a rolling 3 month average, then the following months would also be 33%.  A 4 month rolling average would be obtained if the user entered 25% on each subsequent month.

Another consideration is that if the user just finished a really big month, with record consumption (an presumably sales), and doesn't want the blip to cause excessive forecasted EAU, they might enter a 20% factor instead of 25% or 33%.  On the other hand, if the user has a very slow month, and doesn't expect it to continue, they might want to give a higher rating to that month, say 35% instead of 25% or 33%.  This would have the effect of not understating the expected EAU.