The core earnings of a retail banking operation is net interest margin - essentially the spread between what the bank pays depositors and is paid on loans. That spread has been in a pretty negative downtrend http://research.stlouisfed.org/fred2/series/USNIM/. However, that is for all banks (think BofA as well as you local 1st national whatever), so my guess it may look worse for purely retail operations (maybe not).
My point is this, most figure it costs ~100-150bps to manage a retail branch network (which is a relatively fixed cost) and for the most part (with the past couple of years being the exception) you're getting squeezed in the interest rate side of things (not to mention what you're reserving for defaults). So of course if you are managing one of these banks you are going to want to diversify your revenue stream towards fees.
So why not move to the online only banking model and scrap the branch network? Because retail branch deposits are sticky and are good at gathering stable deposits. That's a much longer discussion - but while some of this may seem idiotic from the outside there is a method (or least logic) to the madness...
Regulation and a less friendly lending environment showed us who's swimming naked with over-extended branch networks: everyone
Both are relatively new entrants to the market, but their account terms/fees are so poor that I cannot imagine why anyone would bank with them. (And judging by foot traffic, nobody is) So why do these branches exist?
Retail banking, on the other hand, I totally agree with you. Basically everything that can be done in a branch can be done on your phone, faster. And the stuff that can't be done on the app can be done by calling in, which still takes a fraction of the time of going to the branch. I've personally had great experience with bank customer service, and routinely have 5 minute calls with them, whether it's to authorize a charge on my card or change my ATM limit.
Banks definitely see this change coming. It's just a question of when they draw the line between catering to the subset of their customers who prefer in-person interaction and the cost of running a branch (especially when compared to the kind of overhead that comes with running an app).
I'm more curious to see how ATM's evolve.
That said, having a bunch of physical branches provides as much an advantage as not having to buy gas when you use a horse instead of a car.
The services around time deposits (providing a checking account, processing transactions, balance inquiries, ATMs) are not that expensive to provide relative to the profits made lending that money back out.
The reason for the explosion of fees is that retail banking in America, as an industry, lost its mind in the 2000s. Every bank brand massively increased its numbers of branches while simultaneously marketing ever more elaborate "private banking" services. Branches are expensive. Marketing to the 1% is expensive. Actual banking is cheap.
This expansion was all funded by cross-subsidy: Banks used to charge huge fees to a small number of consumers who got caught in various traps, e.g. "overdraft protection." A small number of people subsidized services to everyone else. Regulation made a lot of these abusive "services" into opt-in measures, so now retail banks are scrabbling to find new revenue sources to support their bloated, overwrought retail bases.
A more rational system will see the number of retail bank branches fall precipitously. Bank firms that have fewer, more central branches will earn higher profits with fewer fees. Consumers are likely to flee to low-cost offerings. It will take time, but if nothing else, new accounts won't be opened with high-cost banks as often as low-cost banks.
You can already get a preview of the world of tomorrow (and yesteryear) -- credit unions. Small numbers of branches. Lots of ATM sharing. Low fees.