Interesting point, I hadn't considered this.

I would propose that in this scenario the fault is more with you using bitpay than anything else. Sounds like their service doesn't exactly fit your needs. Given the current state of services (bitpay, mtgox) - I think the most forward thinking thing here would be designing your own payment processing software to accept bitcoin as you would then have control over how it works and what features you need / want. Assuming that you could find a decent programmer to design this for you...

...the upside is, if you could manage to develop this, I bet you could resell on some basis to other people wanting to accept micro payments and thus fill a void in the market relative to what bitpay is providing (aka a vpos system where traditional merchants, rather than micro tx merchants can simply sign up and accept bitcoin). Maybe you will become the defacto 'pay with bitcoin' for micro transactions and customers could fund their accounts with you in large amount while you dole out micro payments through more traditional means to other merchants.

~

The other option would be simply using the pre-paid in higher dollar amounts, cutting bitpay out of it entirely and using your wallet software to manually cash out your bitcoins regularly. Not as attractive as it does expose you to some market risks re:exchange rate, but probably only needs minimal software design. Just a couple of scripts: 1 to generate a unique payment address for users to send btc to which is then pushed to your wallet, and another to verify that the address has been funded with correct number of btc.

Then your only issue is holding bitcoins for some period of time before selling them on an exchange somewhere.

If all that is too much trouble to setup/manage then you're back to a turnkey solution like bitpay.

Personally, I'd start with the web side accepting 5 and 10 dollar loads (or however much) and announce that you're working on an easier/better alternative... then go ahead and develop the backend software you need to do it right for your business model. Then work work work to make yourself into 'the micro payment enabling service'

~ (edited to add the following section)

The other option is to see if there's anything the dev's are willing to do relative to enabling micro-payments. This may coincide nicely with some other things that imo need to happen soon for bitcoin. The first is we need to drastically increase the amount of transactions the network is able to process per second in order to remain stable if mainstream adoption does ever occur. Having the tx rate the bankwire system is all well and good (and this is where we are now) but bitcoin is becoming more than that. Ideally I'd like to see us able to process thousands of transactions per second (visa for example does ~2000 per sec) so that growth doesn't make bitcoin unattractive.

It seems to me that one possible solution would be quite simple on the dev side of things. Just change the tx fee option to require only the data size be under a certian limit (we have this already) and that one rather than all of the output addresses receive more than 0.01 - this would allow small micro payments as long as the original address help more then whatever micro tx was being sent + 0.01btc, with the larger amount going to change address.