Basically you're extrapolating future sales that you expect will come from these 50,000 customers. So it's really a question of assumptions + existing data.

Knowing the average sale amount of $120 is nice, but it's really inadequate. How saturated are those customers in terms of ongoing purchases from you? Will they return? Will they respond to upsells or bring their social networks with them? Will the future behave like the past?

"Some customers return to buy widgets". How many customers and which? Which widgets at what cost? You mention a 6-24 month cycle and a $10 to $5000 price range. But there's a huge difference between someone buying a $5000 product every 6 months and someone buying a $10 product every 2 years.

I'd advise against any quantitative measure that simply gives you a handy single number. The topic ought to framed in terms of a range of possible outcomes with margins of error ... as a probabilistic confidence interval. X% chance that values will fall within range Y to Z, in other words.

Undoubtedly, business gurus have their thumb rules, which I'm not familiar with. My background is in mathematical modeling and database architecture; so I can only say how I'd tackle a financial forecast. Those assumptions really depend not only on your data but on an interpretation of your business model and speculation about customer psychology.

Long story short, avoid the "ballpark number" you think you want – at least, if having a reliable number matters.