Incoming long, rambling post that is only somewhat related!
I think there'd be decent cash in creating an organization that accurately rates the size of hosting companies. Investor types prospecting hosting companies to buy would eat it up (obviously they get real numbers once they've found a prospect.) Midsize businesses might even be willing to throw a few bucks at it Angie's List style. Hosting companies are NOTORIOUS for lying and embellishing about their size. Go to any WHIR event (RIP WHIR events!) and everyone would have you convinced they're the size of Softlayer.
There's a lot of publicly visible data about hosting companies - names behind nameservers, IPs (and how many of which are pingable), some traffic info, etc. There's less-obvious ones that I'll keep to myself as well (mostly because if people know I'm observing, they can alter it.) Most of these metrics are similar to unlocking a technology in an RTS or SimCity. To unlock the "Achievement" of doing your own cPanel licensing, you are a size equal to or greater than $500/month in licensing. Plenty of hosting companies wait well beyond that to go direct with cPanel, but it's all just pieces of the puzzle. And, sure, a company might not be big into slapping cPanel on everything, but that's just one of many, many things you can look at.
To pick a more tangible example, let's talk city size. If I begin describing a city to you by certain metrics, you can get pretty decent at estimating the population. For example, if I say the town has an NFL stadium, we're generally talking 1 million people or larger (obviously outliers like a smaller town like Greenbay having an NFL team exist in this example and in the hosting world.) Greenbay's real size shows as soon as you say 25 elementary schools. The more variables you have, the more it all comes together.
Anyways, I think you can determine a pretty good guess based on visible metrics, achievements, and how sophisticated an operation is. And a lot of it (like domains behind a name server) can be automated. And by weighting all the variables with a formula, I think size can be determined in an automated fashion as well.
- VPS, Colocation, and Dedicated servers
Los Angeles, Denver, Chicago, Amsterdam
█ Have your own ASN? Let us scrub your traffic. We have the best DDOS protection.