Menu Close
Blog: Company
August 11, 2010
by Allen Goodman

How We're Building a Better Bank: Negotiating Privacy & Utility


We’re a data-driven company. We believe techniques like data mining and machine learning should be used, transparently, for the benefit of our customers. For us, data mining includes the extrapolation of financial behavior from customer transactions, and machine learning entails the methods that allows us to respond to that behavior based on data we’ve accumulated. It’s worth discussing our process and product, and how it relates to customer privacy.

From my conversations with people interested in switching to BankSimple, I’ve learned that while most people take their privacy extremely seriously, many are willing to compromise a degree of privacy for a more useful product. So when we use your information, it should be perceptible and acceptable to you. And at no point will BankSimple share your information with third-parties without your explicit permission.

Out in the Open

On Saturday, Nicholas Carr remarked in an essay for The Wall Street Journal that “The greatest danger posed by the continuing erosion of personal privacy is that it may lead us as a society to devalue the concept of privacy, to see it as outdated and unimportant.”

Carr reminisces about Scott McNealy’s decree from 1999: “You have zero privacy.” It’s interesting that the recent debate about Facebook shares some of the same characteristics of the reaction to McNealy’s speech. It’s both partisan and contentious. Some believe whole-heatedly that privacy is an antiquated concept. Others see their privacy as a fundamental right that should never be breached.

But Carr is appropriately pragmatic.

He contemplates that “information could be used to influence our behavior or even our thoughts in ways that might be invisible to us.” But, he concedes, “most of us view personalization and privacy as desirable things, and we understand that enjoying more of one means giving up some of the other. To have goods, services and promotions tailored to our personal circumstances and desires, we need to divulge information about ourselves to corporations, governments or other outsiders.”

Furthermore, he explains that the trade-offs each of us frequently, and perhaps, unknowingly encounter online are asymmetric. Most customers have incomplete knowledge about how their information is being used.

We share Carr’s sentiment, so we’re interested in finding the balance where customers are comfortable, aware, and assisted by means of the personal information that they are willing to let us use. We’re using some of the techniques that Carr describes, but only when the customer is cognizant.

In the Conversation

Safe to spend balance

We think it’s in your interest as a customer to provide us relevant behavioral information. This might include how much you’ve deposited and how much you’ve withdrawn. Or, what you’ve spent and where you’ve spent it. We’d look for patterns, then use this information to provide a better service.

Our Safe to Spend feature is an example of how our use of customer information could result in benefits to your life and bank account. I’m excited about this feature because our approach to saving is not one-size-fits-all. I also think this feature demonstrates how we’re clear in our use of our customer’s information. When customers log-into their online bank, they’re greeted by their account balance. What does this balance reflect?

Most banks provide no indication of how their account balances are calculated. This might sound trivial, or inconsequential. But it’s not. In fact, this is a frequent customer complaint.

For example, suppose you purchased Nicholas Carr’s The Shallows on Amazon. Your book might have been shipped, or even arrived, before your bank adjusts your balance. Therefore your actual balance is less than you think. I think most people see how this might benefit the bank rather than the customer. That is, it’s better for the bank if you think you have more money than you do. Whether in spendy behavior or overdraft fees, there’s a reward in it for them.

Better banks give a more accurate view of your account, showing the account balance, pending balance, and available balance. It may also provide a separate transaction history for pending purchases. This is an O.K. solution, but it has significant utility problems.

From a usability perspective, two separate transaction histories don’t help. From a budgeting perspective, I’d much prefer if my purchases were integrated into a single comprehensible transaction history . As far as I’m concerned, once I’ve swiped my card, or pressed submit on the order page, that money is no longer mine.

This method also fails to recognize that some of that money is spoken for. For example, if I sock away a few hundred dollars a month for a vacation I’m planning to take next year, I certainly don’t want to undo the hard work of saving simply because the money is in a pot labeled “available balance.” When all the dollars are unqualified, it become difficult to tell them apart.

From a utility perspective, the standard approach is constrained. Safe to Spend, on the other hand, makes a smart and conservative estimation. Our calculation won’t limit you with an incomplete or at-the-moment balance. It will forecast too.

This feature demonstrates how we’ll be using customer information. Safe to Spend is your account balance minus what you’ve saved toward goals, minus pending contributions toward goals, minus pending bills in the current pay cycle. Therefore your Safe to Spend balance is the amount you’re most likely interested in: how much could you spend without disrupting your budgeting, planning, and ability to pay bills?

Is there a best method to determine what’s O.K. to assume, and what’s not? I think a compromise exists by handling this on a per-customer basis using privacy that scales. We know our customers differ in their financial behavior. So it’s safe to assume they differ in their approach to privacy.

This is easier said than done, it’s a problem that includes a subset of smaller engineering and usability problems, each of which will be a challenge. If this is something that interests you, or you believe you have some solutions, we’d love to hear from you – we’re hiring.