I apologize for not answering this sooner. This is not a bug. The two math calculations performed (an average of three months' tickets and the average of 3 1-month averages) are not the same calculation and will usually result in different answers.
Here is an extreme example of the same scenario to make the difference obvious:
Month 1 = 1 ticket w/ 1 Within Target = 100% Within Target
Month 2 = 1 ticket w/ 1 Within Target = 100% Within Target
Month 3 = 98 tickets w/ 1 Within Target = 1% Within Target
So, evaluated separately, Month 1 is 100%, Month 2 is 100% and Month 3 is 1%.
When you have Xtraction calculate a single percentage for the 3-month span, it is calculating 100 tickets w/ 3 tickets within Target (total tickets for 3 months and total tickets within target for the 3 months).
In this example, Xtraction would return 3% in the scorecard.
When double-checking the number, you took the average of the averages (100% + 100% + 1%) / 3 = 67%.
The problem with taking the average of the averages is that you give equal weight to each month's average, but the months have differing numbers of tickets. The 1 ticket in Month 1 has as much impact on the average as all 98 tickets in Month 3 combined. This leads to the 67% answer. When you group all three month's numbers together (total tickets and total tickets within target), each ticket has equal effect on the average.
Hopefully, this was helpful in explaining the problem. Let me know if you have any further questions.