Simple Post-hoc Sample Size Analysis-for two means

I have a simple problem. I've run some preliminary tests (t-tests) comparing two samples. I was unable to detect any significant difference between the two means; my sample size was relatively small.
I'd like to find by how much I should increase my sample size to detect a difference.
I was advised by someone that I can do this fairly quickly and easy on MS excel by calculating the standard error and changing by N.
I'm sure this is something fairly simple but I'm confused and missing it.
Can anyone offer advice to doing this?
Is there another way I can do this?