A spiritually similar example: Appointment Reminder didn't actually exist in summer 2010, but I had a two-page demo of it set up. I got $400 out of an ATM when I went home to Chicago to visit, and just wandered around the Gold Coast/Magnificent Mile region of the city looking for every hair salon and massage therapy practice I could find. I asked them all if I they took walk-ins and, if so, could I have 30 minutes of the owner's time for whatever the rate was ($30 or so). In lieu of the shoulder massage/etc, I said "I'm interested in the massage therapy industry. Would you mind if we just chatted for half an hour about it?" And I asked about how they handled scheduling, appointments, no-shows, etc etc. I also did a demo of my two-page AR mini-app on the iPad and asked if they would be interested in buying it when it was ready. I think only one person actually accepted my money for the interviews. I got five-ish "Please tell me when that is ready" out of a dozen or so conversations. No Bay Area or signup form required. (I put their emails in a paper notebook. And lost it prior to launch. Whoopsie.) This was mostly successful for me: it confirmed that there was a market willing to pay for AR without me needing to actually build it to demonstrate that. (My sampling technique, which found only massage therapists/hair salons, did sort of lead me off the rails as to who I'd eventually end up targeting for most of the business. D'oh.)
Drive initial traffic to test via PPC in order to avoid social bias.
Or are you just getting them to click buy, only to arrive at a "Not yet open for business" shopping cart?
The polite thing to do is say, "Great! That product isn't ready yet. We'll contact you when it is. What's your email?"
Or you could say, "Hey, we're still working on this. Could we contact you to find out what's most important to you about this product?"
Another option is just to flash an error message and redirect them back to Facebook or Google or wherever they came from. They will pretty much instantly forget about you: things are broken all the time on the web. That's kinda lame, though, so I'd generally lean for something that invites further engagement.