Why is my schedule search produce less result than the same query running adhoc?

[phamxuantung]

Hello,
I have a simple query that run on the last 10 days of month, around 300k events something like:

index=myindex RESPCODE=0
|bin span=1mon _time
|eval length=len(POS_ENTRY_CODE)
|search POS_ENTRY_CODE=8* AND length=2
|stats count

The schedule return 6, which a little bit low than what I expected, so I hit Run to run the same query in ad-hoc, and it return 12, which is right. The time it took to run the schedule was 17 min and the time it took to run the query in adhoc was 19 min. My Splunk version is 7.2.6.

So what wrong with my schedule alert and what should I do to mitigate this problem?

[isoutamo]

What job inspector shows for both queries?