Transfers are killing transit revenue in the Twin Cities.
Metro Transit's local bus routes and the Hiawatha Line only generate about $0.87 per passenger trip, out of standard fares of $2.25 (peak) and $1.75 (off-peak). You might think that heavily-discounted passes are the root of this problem, but this appears to be wrong. If transfers didn't exist, even passengers with deep discounts up to 2/3rds would have to be in the vast majority in order to drag the average fare down to $0.87—a 5:1 ratio or higher—but they are in the minority today. Most riders only get discounts of around 10%, if any.
On the other hand, a single transfer effectively cuts a fare in half, and anyone can do it.
I also feel that the current fare structure discourages short trips. Nationwide, the average full linked trip distance is about five miles. Applied to a $2.25 peak fare, that would result in an average cost of $0.45 per mile. However, passengers going shorter distances get gouged. My own commute is about 2.5 miles or so and only about 1.5 miles of that is on the bus, so I'm paying nearly $1.50 per mile (er, not quite since I have a Go-To card, but you get the drift). By many or most per-mile measures out there, it would appear to be more expensive to ride the bus than to own a car and drive it that short distance (even though in reality it isn't).
I propose that Metro Transit eliminate transfers and instead reduce the individual trip cost to $1.25 or $1.00, which would still be greater than the $0.87 generated today (and would still have a small increase even with a 10% discount). The drastically reduced face value would attract many more riders, since most people only think of the initial cost rather than the full cost when making decisions about whether to purchase something. It would also encourage shorter trips, at least along routes with decent frequency (though it's hard to say if any of those will exist much longer).
With passengers increasingly switching to electronic fare cards, it seems that the value of transfers to both transit riders and transit operators is fading anyway. They could be considered a cost-saving measure, but since every tap of an electronic pass requires a computer transaction, I would think that the backend cost is about the same whether you're looking at a transfer or a new charge to the card. It could be that a transfer is less computationally-intensive somehow, but I can't imagine it's enough to be worth it.
I'd also like to see fares change to a flat rate throughout the day. Currently, about 60% of the transit fleet in the Twin Cities is devoted to the rush hour, and that's a disparity which is probably encouraged by the higher fares that can be charged during peak periods. It doesn't make sense to charge more when a resource is relatively plentiful. (Viewed the other way, it does make sense to discount fares when buses are few and far between, which leads to distress among users. However, I think most people view rush-hour fares as costing extra, rather than looking at non-rush fares and considering them discounted.)
And just FYI:
In 2008, the average Metro Transit local bus cost $112 to operate per revenue hour and carried 36 passengers in that amount of time. Hiawatha light-rail trains cost $176 to operate per hour, and carried 76 passengers/hour. (Those services had operating ratios of 0.28 and 0.38, respectively.)