leveraging logistic results

#1
Here are the results of a logistic regression

Number of Observations Used 97091

Response Profile
Ordered
Value buy Total
Frequency
1 0 75484
2 1 21607

Prob
Level Correct Incorrect
Event Non-Event Event Non-Event

0.5 73141 2932 18675 2343
0.65 67934 6678 14929 7550
0.7 63988 8619 12988 11496

If I have a sample of 30k customers and based on the above I flag (at a cuttof 0.7) (12988+63988)/ 97091=79% of them as events, I am left with 21% as non- event. Can I say that if I only send those a mail piec, from the flagged non event (30000*21/100)=6300 will have in actual of ( 11496/(8619+11496)=57% non event
If I do send all the 30,000 mail piece, the actual non event proportion would be at 23%

Does this is make any sense?