Ameritrade: Hand in Hand

Ameritrade: Hand in Hand

John McCormick

Since the bursting of the Internet bubble, online brokerages have been buying and selling each other like over-caffeinated day traders.

Ameritrade, the largest Internet broker, has been one of the most aggressive. It has picked up National Discount Brokers, mydiscountbroker, and the online accounts of Brokerage America. But its biggest, and riskiest deal, was its takeover of rival Datek.

The 2002 purchase cost $1.29 billion and increased Ameritrade’s number of accounts by 50%, to 3 million from 2 million.

Growth meant survival. But growth also can kill. Unless Ameritrade could meld its systems effectively with Datek’s, it might not have survived the collapse of the online trading business. “The short-term goal was to get the [information-technology integration] done—otherwise the company was going to sink,” says Asiff Hirji, Ameritrade’s chief information officer.

Computer and communication systems are the engine behind any online brokerage, handling everything: taking orders, making trades, and keeping track of client accounts. But Ameritrade went from earning $1.06 on every $10 commission it charged traders in 2000, to a loss of 80 cents per commission in 2001. Absorbing Datek meant $376.4 million a year of additional business. But unless Ameritrade slashed costs at the same time, it could have foundered.

The technical takeover would not be simple. For starters, the company was without a chief information officer. Instead, its chief financial officer, Randy MacDonald, had to keep the integration on track instead and at the same time find a top technical executive to run the show.

Second, the only thing similar about Ameritrade and Datek was they were in the same business. Otherwise, “the two companies were different culturally, technologically, philosophically,” says Hirji, a Bain & Co. information-technology consultant who joined Ameritrade as CIO seven months after the merger.

Ameritrade is a 27-year-old suit-and-tie financial services company; Datek was a dot-com start-up staffed by the sandals-shorts-Starbucks set. Ameritrade ran its business on Sun servers; Datek ran its operations on Intel-based, Windows PCs. Ameritrade was concerned with running a reliable and steady operation; Datek was interested in developing innovative tools for stock traders. This included Streamer, which allows customers to have a wide range of quotes displayed continuously on their screens, along with charts and trading-activity reports.

All this had to be meshed, without a leader of the technology staff. Right after the merger was announced, Ameritrade CIO Raymond Dury left and joined Barclays Capital as one of that bank’s top technology executives. Stepping into the breach was MacDonald, who would later note “the difficulty I had was, I have difficulty turning on my computer every morning.”

The combined company functioned with interim information chiefs for the first seven months. But doing any kind of merger without an established CIO makes it harder to keep staff working on the right tasks at the right pace and tougher to make long-term decisions, Hirji says.

And there were many decisions to be made. One reason Ameritrade eyed Datek in the first place was its innovative approach to online trading. Datek had built a Windows-based system that allowed it to develop new online-trading tools quickly and speed orders to stock exchanges. Ameritrade wanted to bolt the best of Datek onto its computing platform and forge an even stronger information technology framework.

But how do you decide what systems stay and which ones go, without a CIO?

You decide based on what’s good for customers—and shareholders, says MacDonald, a CFO with limited technology background. “Architecture is not something I tried to direct,” he says.

MacDonald boiled decisions down to quantifiable comparisons of performance, functionality and cost. He called in Bain & Co. to evaluate the systems, set priorities and create teams to make the integration happen.

The three key pieces of technology for both Datek and Ameritrade were: the Web sites, which take customer instructions; the back-office systems, which coordinate applications and databases to verify the identity of customers, take their orders and execute the transactions with the right stock exchange; and the internal clearing systems, which make sure customer accounts are up to date.

One by one, Bain consultants evaluated each system. Their charge: rational, not emotional, assessments. It would be too hard, MacDonald felt, for Ameritrade’s staff to kill any of its own offspring. Nobody on a technology staff wants to hear “their kids are ugly,” he contends.

When the results were in, Ameritrade decided to stick with its own clearing system, use some parts of each company’s systems for handling orders, and create a new Web site from scratch that would feature many of the innovative tools Datek had developed, such as Streamer.

The choices extended to personnel as well. A staff of 600 at the time of the merger was winnowed, by merit, to 380. Of those, 100 came from Datek.

“In the technology world, this is not your classic merger where the acquirer nukes 90% of the staff of the acquired company and basically takes it over. That’s not what happened here,” says Hirji.

Taking Stock

The trimmed team worked from the back forward, consolidating the clearing of transactions onto Ameritrade’s system first.

“In the end, the clearing systems were a wash. They were equally modular, they had equal cost performance, equal capacity,” says Hirji. “So, from a risk-mitigation point of view, you’d rather move a million [accounts] than move two [million]” over to a new system.

Moving Datek accounts into Ameritrade’s databases for clearing was relatively straightforward. A procedure was created that duplicated Datek’s data sending and receiving process on the Ameritrade clearing platform so that files could be easily sent between the two systems.

Melding ordering systems wasn’t as easy.

Both companies’ systems consisted of two pieces—an order manager, which handles the customer’s request, and an order router, which communicates that request to the appropriate stock exchange.

The team decided to keep the Ameritrade order manager, since it was more robust at handling trades.

But it found that the Ameritrade and Datek order routers were equally strong and decided to keep both. However, the team felt that the Datek order router was faster and better at handling certain functions, such as trailing stops—a price point set by the trader to automatically initiate the sale of a stock—and plans to consolidate order routing on the Datek system.

The combination of order manager and order routers presented a problem: The Ameritrade and Datek systems could not reliably recognize an order as it passed between the systems. The answer: an electronic token that is now attached to each order, for tracing.

“It wasn’t exactly easy to unplug [the Datek order router] and plug it into the Ameritrade system,” Hirji says. “[But] that’s what we did. We went through that pain,” knowing the improved order-processing speed could be a competitive advantage. This fall, the company expects to start promoting five-second trades.

With the clearing and order systems combined, the company consolidated data centers, keeping Ameritrade’s main data center in Kansas City and Datek’s in Secaucus, N.J. Scuttled: a backup Ameritrade facility in Omaha. The remaining two centers would back up each other.

Another loose end was the empty CIO seat. Ameritrade’s management was looking for a person who understood both business and technology and knew what it was trying to do with the Datek merger. It found Hirji, who accepted the job in April 2003. David Shpilberg, who heads Bain’s information-technology practice, says the company had mixed feelings about letting Hirji go. But it wouldn’t be the first time a CIO was hired from a consultancy brought in to handle a critical technology change.

By the time the final big task—unifying the Web sites—came to the front, Hirji was on board.

But making all the trading tools from both sites available to all customers of a unified site proved problematic.

Datek’s Streamer was built to run on Windows servers, not servers, such as Ameritrade’s, that use Solaris, Sun’s version of the Unix operating system. To adapt the constant-quote service for use on its servers, Ameritrade used a Java plug-in to provide a link back to Datek’s Streamer servers.

But the plug-in didn’t always work due to a fault in the program. As a work-around, Ameritrade added a Web server to the system that allows traders to link directly to Streamer.

Streamer may now work, but Ameritrade’s site is still considered lacking in tools and not easy to use, according to Gomez Inc., which monitors and ranks Web sites by cost, customer satisfaction and other parameters.

Phylis Esposito, Ameritrade’s chief strategy officer, notes that Ameritrade ranks well on its ability to execute transactions, which was the company’s first priority. Only recently has the company been able to turn its attention to making the site easier to use and navigate. Coming: the ability to initiate a trade from any page on the site. Hirji calls this a “persistent trade ticket” and says it will appear when the new Web site rolls out, most likely before the end of 2003.

The Final Line

Despite the technical challenges, business goals are being met. The two companies’ systems have been integrated into a single processing platform that can handle nearly half a million trades a day; currently it’s handling 158,000 a day. The company cut $80 million in technology-related expenses in the consolidation, allowing it to beat its overall cost-cutting target of $173 million by more than $70 million. And the company hardly has to worry about another collapse of its market. Its break-even point—the amount of activity at which it can start to make money—has been reduced to 29,000 trades a day, from 50,000.

This is reflected in its bottom line. On Oct. 21, Ameritrade posted record year-end results—net income of $137 million on revenue of $713 million.

Not bad for a company that went from a profit of $69.6 million on revenue of $654.5 million in fiscal 2000 to a $91.2 million loss on revenue of $498.7 million in fiscal 2001.

It certainly is a big enough swing to make MacDonald a happy man. On Oct. 22, at the “CFO & IT Value” conference in Chicago, he crowed that Ameritrade’s results now put it in the “95th percentile” of large American companies, in profitability.

And, even though he has left his temporary CIO hat behind, he also said that Ameritrade was not really a financial-services firm, after all.

He said he now considers it to be … a software-development company.

Ameritrade Base Case

Headquarters: 4211 S. 102nd St., Omaha, NE 68127

Phone: (800) 237-8692

Business: Online financial-services company

Chief Information Officer: Asiff Hirji

Financials: Revenue, $713 million; net income, $137 million; net profit margin, 19% of revenue. Figures are for fiscal year ended Sept. 26, 2003.

Challenge: Improve stock-trading features and functions while incorporating the operations of an acqusition, Datek, into its own systems.

Baseline Goals:

Cut $173 million in annual costs, about half of which could come from integrating Datek.

Support 450,000 trades per day. Average number of trades per day in fourth quarter ended Sept. 26, 2003, was 158,000.

Retain more than 90% of customers from both Ameritrade and Datek.

Copyright © 2004 Ziff Davis Media Inc. All Rights Reserved. Originally appearing in Baseline.