20% savings is not ground shaking. Remember that BTC is tackling an entrenched opponent, quasi monopolist. It'd take an order of magnitude in savings to guarantee success. Think 50% to 90% less.
I think you're also misunderstanding of how the financial system looks at payment rails - there's a pent up (ongoing?) demand for better ones.
Also, while merchant transactions are one market, they certainly aren't the only one. There's lots of talk about international remittances. This is a $500B+ market[1] that is growing at an insane rate[2]. Average fees are 9.3%[3] and way worse for some countries - it's not unheard of to say, pay $40 for sending $100 from the US to Kenya. Now that is a market that's ripe for disruption (you'd have to balance bitcoin inflows w/ capital flight for the economy to work, but it's worth pointing out there's already a service linking bitcoin w/ M-Pesa: http://motherboard.vice.com/blog/one-third-of-kenyans-now-ha... )
IMO, this is a particularly useful page which gives some context on where bitcoin currently sits with other payment networks: http://www.coinometrics.com/bitcoin/btix
[1] http://www.bloomberg.com/news/2013-12-17/global-immigrants-s...
[2] http://gulfnews.com/business/economy/global-remittance-flow-...
[3] http://www.cognizant.com/InsightsWhitepapers/Remittance-Mark...
If Amazon were motivated to take on the risk of building the infrastructure, this could happen quickly. But, they have other things to focus on, and I doubt they would make the calculations in the same aggressively optimistic manner.