An easy approach to this would be to create 2 streams in your Engagement Program programmed to cast the same content at the times you're wanting to test. Stream 1 could be 9:00am, while Stream 2 could be something like 9:15am, 8:37am, or whatever interval you think might yield interesting results.
I'd also try to eliminate as many other variables as possible before you test timing of the send, so you can just analyze the effect of time on your audience, instead of trying to weed through other possibilities that might change your results.
Good luck with your program!