Matt McCall, an investor in EIP and good friend (and fellow Kellogg MBA alum!), recently wrote a great blog about how VC's look at businesses to invest in. See Scalability -- P*Q
We run into exactly the same thing at EIP -- how well does a product concept scale? And we recently updated all our product submission and public speaking decks to reflect how we evaluate market sizes for products. We see a lot of inventors that believe their product will take 1% of a big market and is therefore a huge idea. As Matt says, top-down models don't work.
In our world, we ultimately take Matt's model of P*Q down a layer and view it this way: target market size * awareness * distribution * trial * price + [repurchase rate]. That is, how many people are in your target, are aware your product exists, can buy it somewhere, and what % of people indicate that they'll buy it when you do some surveys with random consumers (never trust your friends and relatives). Then we factor in repurchase rates for products that are consumable. This can get highly complex and folks like AcuPoll and BASES will do a wonderful job making highly accurate forecasts.
But, I'm going to give every budding inventor a cheatsheet here. We can simplify the above to X * R * 0.04 where X = number of customers that you think are target customers (US only) and R = price point of your product at the store times the number of times a year the consumer buys it.
I won't bore you with the math, but trust me when I say it works surprisingly well (feel free to email me and I'll send you the details). Using this model, if you can get over $30m a year, we're definitely interested, as is almost any branded manufacturer...