I think they would have been more successful if you could have actually gotten them for the products you wanted, not just a random selection of stuff you probably never ordered.
So when you buy a $5 Dash button for Tide Pods™, Amazon spends $11 to make it and ship it to you, of which you pay $5, Amazon pays $2, and Procter & Gamble pays $4. Amazon and P&G both gamble they'll make up the loss on the button with profit on your extra purchases.
This explains a bunch of weird features:
* The reason Dash buttons were only available for some brands is because only some brands had agreed to pay the subsidy.
* The reason they ship with the logo sticker already in place, instead of a single item with a big sheet of stickers? Because the sticker is for the company that paid the subsidy.
* The reason non-Prime customers couldn't buy them? Because if you don't have free shipping, they think you won't press the button enough to make up their loss.
* The reason the AWS button costs more? Because nobody's subsidising it (plus the fact they charge what the market will bear, obviously)
* Why would they discontinue them? Because it turns out the gamble on extra sales doesn't pay off.
Then again, I agree that most people willing to use a Dash button are also the kind of people willing to use an Echo, save the uber privacy concious minority.