What do you do next? Go to the target business? Let's say you go ahead with this idea, what if half the people drop out if the deal amount is doubled or %off is halved?
I'd like to know why the sliders, and how it's different from Loopt: http://www.businessinsider.com/loopt-u-deals-2011-6
For users, maybe let them also offer non-monetary factors for discount - off-peak / non-prime hours, introduce a friend, available on short notice to fill a cancellation, post a blog or tweet, post a local sign, ??, etc. - bartering a little for the discount where the retailer would see value in the barter. Groupon can't do this, but your personalized approach seems ideal for this.
In cases where business finds deals that work, archive them for the business, then offer to post or syndicate them on-demand as desired by business to drum up demand on an ongoing basis.
And to keep users interested, do something where you as the service do random (or not so random) selections and 'make the discount' happen for the requester, at _no_ cost to the merchant.
Based on what I have seen from your "users" (or army of employees) people are requesting deals from businesses that they already like and frequent. Generally speaking, you frequent a business because you like and receive what you perceive to be fair value for what the money you spend there. If I go to the cafe down the street 5 days a week because I like it and am willing to pay full price, I have no real reason to ask them for a discount - I am satisfied paying full price.
Now, it does make sense to me to request discounts from places that I want to try as a way to get me in the door. However, actually requesting a deal from these new places requires work on my end, whereas I can just sit around and lazily let Groupon/LivingSocial push new deals to new places on a daily basis.
Cool idea, but not something I see myself using.
Ultimately it seems you end up in the same place as Groupon as far as using it. People would want to know what deals were already available, so they don't have to ask for one themselves most of the time.
It's got a massive chicken-n-egg problem, so that's the real challenge in making it work. Using the "bowling pin" strategy and starting in Santa Monica should help.
It's interesting and different, which I like.
I think a typical use case would be someone looking for a better deal when they've got a big group of people to entertain, in which case there's a particular time frame they're interested in.
Fix the discount at some reasonable number around the average of Groupon-type discounts.
The amount spent will be decided by the type of business anyway.