For all the press they get, pure Internet banks account for about 10 of the 1,121 transactional Web banking sites that were in place as of mid-December. A phenomenon since late 1995, Internet-only institutions have yet to make a material impact on the banking industry.

Their sluggish performance, overlooked until now, has recently become a source of widespread criticism. This may reflect something of a fashion in the analyst community, but also the fashion in banking whereby traditional banks are establishing Internet-only subsidiaries. That trend has prompted an inquiry into what the pure Internet bank has to offer.

Low share blues

Their share of outstanding loans and deposits is about 0.05%, or five basis points, according to the latest research by A.T. Kearney Inc. (which tracks publicized numbers and does its own analysis of them).

Internet banks' share of online banking activity, ironically, is a low 4%, according to a large consumer survey conducted late last year by Gomez Advisors Inc. The results of the 19,000 interviews were released last month.

Virtual banks account for only about 2% of online banking households, according to an analysis by Mainspring Communications Inc. By contrast, the seven traditional banks with leading Web shares own 63% of the online market, the Cambridge, MA-based firm says. (The Web share of traditional banks was analyzed in last month's cover article, "Remote Banking Rankings.") Mainspring counts the uncontested top five in its top seven list-Wells Fargo & Co., Bank of America Corp., First Union Corp., Citibank, and Bank One Corp.-along with Chase Manhattan Corp. and the pre-merger BankBoston Corp.

A ranking of "pure" Internet banks is harder to do, not least because of the lack of consensus over which entities to count. For instance, Eric Rajendra, an A.T. Kearney partner and head of its global e-commerce practice, distinguishes three types of Internet banks.

- "Cyberbanks" include Net.B nk, Telebank, and CompuBank.

- "Separate Web subsidiaries" include, for instance, Bank One Corp./First USA Corp.'s WingspanBank, or Security First Network Bank, which has been owned by Royal Bank of Canada since late 1998.

- "Proprietary new channels" such as Citi f/i,, and

Rajendra sees value in distinguishing the two types of subsidiaries in that, say, Wingspan is under less parental control than (Note: We use the term cyberbank in a general sense, not Rajendra's specific meaning of the term.) Rajendra adds that he doesn't see excellence in any of these categories, a topic to which we shall return.

Big enough?

The accompanying table ("Pure Internet Banking Leaders") gives an indication of which Internet banks have the largest customer bases and holdings. The number of accounts is higher than the number of customers, since each person typically has between one and two accounts with the bank.

Rajendra's cyberbanks and separate subsidiaries dominate in the table. However, one could argue that it includes all three of the somewhat discretionary categories, given the inclusion of mbanx direct, a site operated by Bank of Montreal. Our ranking is gleaned from minimal direct reporting by the banks (always quoting numbers at the high-end of the range), variously timed analysts' reports, and second-quarter financial information supplied on the Federal Deposit Insurance Corp.'s Web site. Third-quarter results had just been posted on the institutional directory page ( as of late December. Two Canadian virtual banks, included for some perspective, dwarf the U.S. Internet banks.

In the U.S. it's hard to tell whether WingspanBank or Telebank is larger. Wingspan is credited by some with opening 100,000 accounts, while others suggest it has only 15,000 customers. GartnerGroup analyst Laura Starita said she heard that that's how many people who signed up actually provided funding to open their accounts. At this point, it's almost impossible to tell whether Wingspan is a success. Some sources indicate that it's deemed a failure, partly explaining the recent batch of executive departures from First USA and Bank One. On the other hand, several analysts consider it the best, most innovative Web banking effort yet.

Telebank is far less controversial, though some contend that E*Trade paid too much for it-$1.8 billion, based on the initial pricing on June 1. E*Trade, the second-largest online broker, avoided a contractual penalty by closing the acquisition just before year-end 1999. That aside, Telebank is generally credited with having at least 60,000 customers.

However, SFNB, arguing that its customers are all true Web bankers-unlike those of some of its peers-contends it is in the lead with 46,000 customers as of December. Moreover, if one uses more of a qualitative analysis, SFNB would lead, having won the award for best banking site four times in a row from Gomez Advisors Inc. In those ratings 'Net banks generally outshine traditional banks. Furthermore, Bauer Financial Reports Inc. of Coral Gables, FL, recently gave Net.B nk, Telebank, and SFNB, five, four, and three stars, respectively. On that score, however, Rajendra is dismissive. "A cool Web site may result in bankruptcy in two years," he says.

Forrester Research Inc., which does its own consumer research on 100,000 North American households, echoes Mainspring's suggestion that 'Net banks have a 2%, or maybe 3%, share of the online banking market. "However," says Brook Newcomb, a Forrester senior analyst, "if we focus only on start-ups like SFNB and Telebank-and exclude offshoots from traditional firms like Bank One, Citi, and American Express, all of whom have mammoth customer bases to cross-sell to-the number is probably closer to 1%."

Profitable enough?

Worse, the returns of cyberbanks are lower than those of their peers, according to a Mainspring analysis published mid-December. Since most pure Internet banks are thrifts, Mainspring compared a group of traditional and virtual thrifts of roughly equal size, a useful comparison because the Office of Thrift Supervision chartered five of what the firm considers the eight pure 'Net banks. An OTS spokesperson adds that at least another three applications for 'Net thrifts are pending approval.

The results of Mainspring's analysis are shown in the table entitled "Performance, Offline and Online." Although there was considerable variance in their return on assets, for instance, the cyberbanks collectively had an ROA of -0.02%. The relatively small thrifts, by contrast, outperformed the thrift industry's 0.98% average, with an ROA of 1.07%. The cyberthrifts return on equity was -0.18%, compared with their traditional counterparts at 11.64%. Cyberthrifts even earned less on their investments, partly because they play little role in the potentially lucrative mortgage business-a mainstay of most thrifts.

The regular thrifts had assets ranging to $1.4 billion-with $991,121 on average-while the cyberthrifts had assets ranging to $3.2 billion, but $1,052,245 on average. The traditionalists were Brooklyn Federal Savings Bank, First Federal Lincoln Bank, Harbor Federal Savings bank, Home Savings Bank, and York Federal Savings and Loan Association. The newcomers were Net.B nk (established October 1996); Telebank (established November 1997); SFNB (established October 1995); and First Internet Bank of Indiana (established February 1999).

Not that Mainspring suggested time will cure the cyberbanks. Growing their customer bases won't help the cyberbanks' fundamentally flawed business model, the analysts suggest. "[It] won't solve the bigger problem of narrower interest margins caused by high deposit rates and low investment returns," states the December report.

Add branches, cut costs

Although the proverbial pure Internet bank is often described as "branchless," many of the banks that started with only an executive office have opened a couple of branches and big call centers. That includes the original of the species, Atlanta-based SFNB.

Others do more business by phone than online, including Telebank, which, as of the third quarter, appeared to be the largest domestic pure Internet bank. As its name indicates, Telebank's original channel, back in 1989, was over the telephone. Some contend that its main business is still phone-based sales of certificates of deposit. Retail CDs accounted for 75% of the Arlington, VA, bank's deposits last year.

Similarly, Canadian virtual bank ING Direct swore off Web banking entirely when it launched in May 1997, and again a year later. Its then 60,000 customers have grown to a reported number of 250,000-by far the largest of any North American bank, and almost as many customers as all of the leading U.S. pure Internet banks combined. ING Direct, a subsidiary of Dutch financial giant ING Group, succumbed to Web banking last spring, another sign of the pressure on all banks to serve all channels.

The theory was that virtual banks would have lower overheads, allowing them to offer preferential rates on loans and deposits. Practice has shown that not to be the case. That's exemplified in the aforementioned Mainspring analysis and another publicized later in December by A.T. Kearney.

Both help to quantify a September report put out by Meridien analyst Dana Stiffler contending that virtual banks are not viable long term. (Rajendra claims he said as much in mid-1999.) In fact, Meridien expects only 5% of a far larger online banking consumer base to be with 'Net banks come 2003. Meanwhile, Stiffler says, many of today's Internet-only banks will either be bought or "absorbed back into the institutions that spawned them." She notes that that's what effectively happened to mbanx, a virtual Bank of Montreal subsidiary established in late 1996. Though it continues to have its own Web site, technically and operationally it reverted to parental control last August, when it was renamed mbanx direct. It has attracted at least 160,000 customers (according to Meridien's September 1999 research), but that may partly reflect the consolidated Canadian banking market.

Virtual banks have shifted rather than cut their costs, analysts aver. The brand recognition they lose by foregoing physical branches, they overcompensate for in what they spend on marketing and ill-considered Web alliances. The marketing expenses of Telebank, for instance, takes 11 cents out of every interest revenue dollar, notes Mainspring. As such, the bank would need to quadruple its account base to 375,000 to match the ROE of offline competitors, even though Telebank has the lowest personnel and facilities' expenses among the virtual banks analyzed.

Worse, virtual banks have not automated many of the paper-intensive processes that support the Web transactions. As Meridien's Stiffler says, "When Web-savvy that their cyberbank is actually a front, they are not likely to stay long."

Some examples of pseudo-electronic transactions, notes A.T. Kearney's Rajendra, are electronic billing, where hardly any of the consumer's bills are presented electronically, and mortgage origination, where the associated real-estate services (such as insurance and appraisals) are not in the electronic loop.

However, Rajendra says that if anyone did Web banking right, they could afford to pay the typical 1% more on deposits and still reduce their operating costs by between 15% and 41%. That's according to an analysis he conducted recently for a traditional bank contemplating a big move onto the Web. It also assumes that the bank would charge 1% less on its loans. Whether the bank cut costs by 15% or 41% would depend substantially on whether it kept what it saved from a more efficient process or passed on those benefits to its customers. He didn't give concrete examples of how a paragon of Internet banking might operate so as to "reengineer the value chain."

And the winner isn't...

Analysts generally report that no bank site blows them away. In fact, in a discussion at the recent Bank Administration Institute's conference in Miami, Josh Grotsein, the Citigroup executive in charge of e-Citi, said he would grade the banking industry only as a "C+/B-" for innovation-although he added that the focus in 1998-99 was in "getting a platform out there." Wells and Wingspan get widespread credit, the latter for its third-party lending and insurance offerings. is considered the worst major-bank Web effort by one analyst, who preferred not to be named. He also considers Citi f/i still merely a concept, despite its official launch last August.

The viability of virtual banks was something of a talking point at the BAI conference. James McCormick, president of First Manhattan Consulting Group, who questioned the premise that Web banking customers are inherently profitable, noted that Net.B nk, for example, has a net interest margin equal to that of about four regular branches and a deposit base ($490 million) equal to that of nine average branches.

It seems that the Web is proving a challenge for all banks. Even American Express Co., with its famous brand and 22 million customers, is finding the going slow. Francois Odouard, an Amex vice president, said in a panel discussion at the RDS show, "Banking is a very sticky market. It takes a lot to acquire customers, which is not that good news for us."

This article appeared originally in the February 2000 issue of U.S. Banker.

Subscribe Now

Access to authoritative analysis and perspective and our data-driven report series.

14-Day Free Trial

No credit card required. Complete access to articles, breaking news and industry data.