In the context of predicting success it does not make sense to put talent - or other non-changeable initial endowments like country of birth, parental wealth, attractiveness - into the "luck" bucket. While you can't do anything about these factors, an accurate assessment of them will tell you a lot about your chances of success in a given domain.
I.e.: you might need all three of exceptional talent, hard work and luck to succeed in an endeavour - and while luck might be random, and "hard work" can be influenced by yourself, the talent factor can be gauged but cannot be influenced so it can be a hard limit that should guide you in the path you take. I.e. you don't dedicate your life to being a supermodel if you were born butt-ugly, or to be an NBA player if you are 5 feet tall. You might succeed, but your chances are microscopic.
Two people could have equal ability to make decisions - some good, some bad. In some fields of endeavor, they might succeed about equally.
But in startupland, it's easy for one wrong decision to kill the business, so they are removed from ever trying again. Or one right decision could cause them to experience accelerating returns – which compensates for all the other dumb mistakes, and eventually kills off their competitors.
So, paradoxically, the more cutthroat the environment, and the more exponential the possible returns, the less we can use survivorship as evidence of superior skill.
Note that I'm not saying a successful businessperson is unskilled. It's just that it's not conclusive evidence of superior skills.
It's reasonable to assume that becoming successful in business requires you to make more than 20 correct decisions, and depending on the level of success, probably closer to 30-50 good decisions, and a population of billions and trillions just to find one successful person.
Conclusion: real life is not just coin flips. If it were, noone would be successful.