How can banks make their mobile apps more competitive? We recently asked Greg Nudelman, principal and CEO of San Francisco-based DesignCaffeine, who has worked with USAA, Intuit, and Wells Fargo on app design, about his pet peeves.

What are the biggest mistakes all app developers, but especially bank app developers, make?

Too many people try to approach app development the way they approach web development. They say, "How can you simplify the [website] experience to the level that it can be fit to a small screen?" That's the completely wrong question to ask. They're missing the entire opportunity that is presented by devices. You have to start from the ground up. The best way to approach that is lean methodology.

The first question you have to ask [the user] is, "Where are you?" That drives a lot of the use case development. In the past, we've looked at personas first and foremost because, for instance, if a retired grandmother of six and a young lawyer apply for insurance online, their priorities and their paths will be quite different. Personas have been an excellent tool for us to explore people's behavior and create applications people know and love. But when the grandmother of six and the lawyer need a parking space, they have similar behaviors — it's more about your context, not your persona any longer. The context will include the persona. 

So would you say collecting geolocation information and convincing people to opt in to that is critical?

Absolutely. That's just the beginning. The new devices in the Android line and Apple's longer-range Bluetooth devices have shown us the way of the future and how much integration we can expect further out. The Galaxy S4 helps the car respond to the owner by knowing who's sitting in the driver's seat. The NFC chip activates, the seat adjusts to your position and your favorite stations come on the radio. We're just scratching the surface with this capability, where truly our environment is responding to us through this cybernetic sense.

In the financial world, many people are miles behind the curve. The financial life has always been bogged down, it's difficult to get at your data. USAA debuted a four-digit unlock code before many other banks did. That alone made it much more usable than the bulk of the applications out there. And GoBank has an interesting feature where by swiping on the login screen, you can get at your balance information without typing in a password. [Several other banks have this feature, too, including ING Direct, Bank of the West and City Bank Texas.]

What are some ways banks can make mobile banking easier, requiring less data entry on the customer's part but still maintaining security?

We've explored different on-board sensors that are available on the phone, and there's more stuff on these devices than we know what to do with. You can have things like biometric pressure, which allows you to figure out which floor of the building someone's on, proximity sensors, things of that nature.

The current systems, as difficult as they are to access, are not particularly secure. When you think about the four-digit code that a lot of banks and Google Wallet use, that code is what most people use to unlock their phone, their ATM card and their online banking account, and that is a huge breach of security if you think about it. It's very easy to hack and once you crack one you can get access to a person's entire financial life. We've got to find different ways to secure it. I really like Apple's approach with the thumbprints. I also really like Android's approach with facial recognition.

What are some of the shortcomings of the current crop of mobile banking apps?

A classic pattern is the form-first pattern, where you take a web form and convert it into a mobile alternative by presenting a blank form. Unfortunately, when you look at it on a mobile device, each field takes you to a separate screen, for instance to pick an account number from a big list, and then it goes back to the form again, partially filled. It takes eight tabs or screens just to transfer funds from checking to savings. I think there are much better ways to do this. For example, if you take the wizard approach, which provides a dedicated screen for each of the fields, the transfer can be done in four steps, half the clicks. It's the same number of fields and the same amount of data but very different screens and flow.

A year ago I spoke with Brad Smith at Intuit, who was talking about adding more data prepopulation to their apps to save user time and effort. Recently I spoke with AT&T and Verizon, who are trying to offer people a simpler way of accessing their accounts by populating their new account forms with phoneaccount data. Does that idea have legs?

One of my favorite topics is search. Search has turned from being a feature to being integrated into every aspect of our digital lives. Any time you do a look ahead and suggest the completes, it's a valuable feature to bring to the device and one that is very often overlooked. It's a delighter and an accelerator for people. I've applied for a patent on technology called tap-ahead, it's an auto suggest that suggests one word at a time. So as you start typing something into a field and it lists matching one-word terms. If it's a word like "Harry," if you type in "ha" you can have a bunch of matches like hair and Harry Truman, Harry Potter, every Harry there ever was. Very quickly you could explore a space of thousands of suggestions through this matching. It's as efficient as keyword browsing where you explore the space one word at a time. I think we don't do enough of that to assist people in getting data into our devices.

Then using alternatives such as voice command is a valuable thing. If voice command isn't quite perfect, it could come up with a list of suggestions. The entire paradigm of typing things at a keyboard is becoming obsolete when it comes to mobile devices and tablets. Tablets are a little easier to type on, but they're still a pain.

At another level, you have things like Amazon's Mechanical Turk interface for overseas agents, where you scan very complex documents but at the back end, humans armed with informational intelligence and character recognition verify the data. So maybe a computer does a first pass and then somebody checks it against the document visually to make sure the software has captured what the document was doing. That concept can be used in mobile apps. For instance, Card Munch, which is owned by LinkedIn, scans business cards and inputs all the data from them into a contact database. Taxbrain lets people file taxes through an interface that even allows you to speak with an agent avatar. At the back end it's controlled through a human being and a machine working in tandem to help you file taxes. I think that's the future of complex data entry.

I wonder if we'll get to more tailored applications over time, where the first few times you use an app, it learns your preferences?

That's a wonderful idea and something we're starting to scratch the surface of, pulling in suggestions the entire time the user is experiencing the app. Amazon is fantastic when it comes to that, Netflix has built an entire service on movie suggestions, YouTube is a bit behind on a lot of this stuff. But I think Amazon and Netflix are showing us that just from the power of creating good suggestions, you're capable of delivering a very valuable service.