I haven't checked how the magical 3 is arrived at but assuming it's reasonable I think a more valuable exercise would have been to hold the 3 constant and plug

I haven't checked how the magical 3 is arrived at but assuming it's reasonable I think a more valuable exercise would have been to hold the 3 constant and plug in the values you're most sure about and then speculate about the missing variables. For ZocDoc with its 1-10k CAC range you could then speculate if they could reach the CAC that would lead to the magical >3. So basically I think it's more valuable to use this as a tool to check if you think a startup can realistically optimize towards the >3 or if that is futile (see how they can influence LTV and CAC respectively) instead of trying to "rank" startups by plugging in LTV/CAC if you're not sure about them. Interesting read either way.