Very, very hard to determine these days.
First of all you have to understand that Games Journalism these days consists of companies making deals with various reviewers (IGN, Kotaku, Gamespot...) to give them a free review copy and an embargo date (this is the date when sites are allowed to post reviews before launch). A game company being disappointed with a site's review of a previous game would result in strained relationship between the two parties. There was a guy that got fired from his job because he gave an honest review of the game Kane and Lynch. Source (http://www.gamespot.com/forums/system-w ... -29097920/)
Because of this sites are forced to suck developers' ------- before the launch of a game or else they wouldn't be able to put out a review and would lose site traffic on launch day. So even if a game is bad and riddled with bugs, there's a good chance sites would still give them 10/10 reviews.
Now for mobile I don't know the exact parameters.
I mean you could create Earwax and Snot and Undigested-Peas-and-Corn Kernel-clasts in-a-Turd versions of Candy Crush today and wake up a billionare tomorrow. (which itself is a clone of Bejeweled which itself is a clone of a Russian game called Shariki)
Take Flappy Bird for example, it's not great, it's good mechanics-wise; does what it's supposed to do but somehow it generated a shit ton of money.
I'd have to say the first thing you have to consider is learning curve. Flappy bird has zero learning curve just to play the game but there's the challenge of getting good AND getting consistently high scores.
Start from creating a game that has virtually zero learning curve but gets progressively harder. If you're getting positive feedback from friends and random people then you only need better art and marketing to turn it into a success story.