So, for any online merchant, Amazon is worth a look and an experiment. It can give you INSTANT access to a vast buyer base and drive a TON of sales. Of course, this has it's own benefits - not only will your volumes increase (enabling you to push your suppliers for better discounts) but you also are accessing buyers that probably wouldn't have found your business or your products otherwise - which means incremental sales over what you would do if you didn't sell on Amazon.
Your points are all well thought out, and very valid. However, one very important point you don't mention is actual retail price. I'll illustrate this with a real life example.
We run a number of retail sites, and sell exclusively through these, and not on third party channels such as Amazon and eBay.
One of these sites specialises in a particular type of household goods, with an average sale price of £50-£100. Typical margin for this type of product, when retailing at RRP, if 40% gross (that's the best achievable margin).
One of our main competitors do sell on Amazon. They typically sell at 75% of RRP. So something which should sell at £100 (and does on their retail site), they sell on Amazon at £75. This is necessary in order to compete with other Amazon sellers, and with Amazon themselves.
So instead of having a £40 margin on this product, they have a £15 margin. £10 of this goes to Amazon. You are then left with £5 margin on a £100 product.
This is not sufficient margin to justify the sale - once you've taken your running costs out of the equation, you lose around £20 on that sale. So the sale is only worth it if you have a near 100% rate of customer retention. This is never the case, especially not with customers buying from Amazon sellers, who invariably shop on price.
The ease of accessibility (for new merchants) of eBay and Amazon is a double edged sword. On the one hand, it levels the playing field. On the other hand it encourages swathes of new, naive sellers, who have little or no understanding of the underlying numbers, to have a go at selling at minimal margins, margins which mean they actually lose money, and stop trading within 6-12 months.
We've looked into this very closely, and concluded that selling on Amazon could indeed generate huge turnover for us, but at an operating loss. It would in effect be subsidised by the core business. This applies not only to the product discussed above, but to several classes of consumer products we looked at.
If it were possible to sell on Amazon at RRP, with 15% fees, we would.