The numbers don't add up!!

Yes, I hear you there ... and your comment about the inconspicuous nature of the opt-in did give me some pause for thought. However, there are other possibilities, too? Ray's point above, about hosting being a "recurring" thing is a good one, for a start? And there's also even the possibility that someone very inexperienced is testing something and won't be doing it for long? (In general, it doesn't pay to assume that something's necessarily profitable, just because you see someone apparently doing it?).

.

Mike's on the money $$$

Now you see, that I'm not really in the affiliate game , I've just worked the numbers, and assumed, that EPC means (average) earnings per click, and calculated with that. Of course per 100 clicks makes the math to add up harder, but it can still work.

The EPC actually suggests that about 1 of every 200 visitors convert on the affiliate partner's site (we still don't know how many visitors the toptenbest....com site itself converts). The EPC is around $40, and the flat rate commision is $90-$100 for one conversion, so only 0,5 out of 100 visitors generate commission.

There are still two important factors:
The EPC is a CJ-wide average, and contains statistics form a lot of low-performing affiliate partners as well. Many of those partners have websites or lists in quite unrelated, or just slightly related topics, and just give hosting affiliate offers a try to generate a few bucks.
This is not the case however with toptenbest...com. They are very-very relevant to the actual affiliate offer, and also their keywords are very relevant. If someone searches for hosting related terms, they most probably look for some kind of hosting. For someone like this a toptenbest....com-like review and comparison site is probably the best option to compare the offers, so a huge percentage of visitors will go through the offers and convert eventually.
And from toptenbest....com's point it doesn't really matter whether the visitor goes with the 1st or the 10th option in the list, as long as they check them one-by-one, and come back to toptenbest...com, to check the next (by clicking on another affiliate link). They will still mean one visitor-one conversion for toptenbest....com, just will have more pageviews and more time spent on the website (but they probably don't really care about that).

If this website can actually convert let's say 1 out of every 3 visitor for one of the affiliate offers on the site, than with 90$ flat-rate commisions, it would mean:
$90/3=$30 average commision for one clickthrough, so their own EPC would be like 30*100= $3000, if my calculation is correct.

There's also the question of CPC:
As @drewfioravanti said, the actual CPC costs could be very very far from the CPC suggestion of keyword planner, especially in case of such high-spending and good-performing advertisers as toptenbest...com. I assume, that they spend a lot, and perform good, because their ad is in the 1st place for a quite competitive keyword.

Google is great in understanding how it really makes revenue from advertising, and thus favours those with older, high-spending, good performing (meaing quality score and CTR) accounts over others.

Google's revenue is clicks*bids. If someone with a poor ad but a high bid of $10 would get in the first position, and would generate say 1 click out of 1000 impressions, Google would get $10 revenue.
If someone with great performance, great quality ad (which they can determine by historical data and other factors), but a much lower bid of $0,5 would get to the 1st position, but would generate 10% CTR, meaning 100 clicks for 1000 impressions, Google would still get $50 revenue, 5 times as much as with the other guy, with the 20x higher bid.

Just as I said earlier, a lot of info is missing and too much guesswork is involved