So first, let's look at examples in the banking and the financial industry. These kinds of examples tend to focus more on fraud detection, credit and risk, personalized banking, regulatory compliance, e-banking conversion, converting customers from traditional banking to e-banking, they'll focus on lending and investments. Very often though involve credit card usage, ATM usage, and understanding of mortgaged data, securities ratings, often called surveillance, venture, hedge funds, MNA assessments, and brokerage type solutions. Let's take a look at a company called Vantiv. They're a payment processing company that wanted to leverage its expanding big-data lake and related resources to expand revenue opportunities. They used an analytic solution to visualize and populate domain-specific objects from their data lake. Then they mined the merchant data to develop precise pricing models. The new pricing models and ability to conduct what-if analysis on large datasets, helped them open additional revenue opportunities by repricing key customers. The results were, that they were able to boost annual revenue by eight to $15 million per year by more intelligently pricing customers, and in the process they save $2.5 million by eliminating hard copy financial statements. Next, let's take a look at a Russian bank called Tinkoff Bank. They wanted to overcome challenges with integrating, storing, and processing big data for improved customer intelligence. So they integrated a 150 terabytes of semi-structured and unstructured data, from the real-time billing platform, their mobile applications and website plus structured data, from various databases into a data lake. The results were, that they got a complete near real-time view of each prospects credit risk profile to determine credit viability, resulting in a tenfold increase in marketing upsell campaign conversion rates. Commonwealth Bank needed to detect a range of financial and operational fraud, via a single platform. The resulting solution enabled them to improve check fraud detection by 95 percent, and fraud alerts by 60 percent, as well their internet fraud loss-to-turnover ratios improved by 80 percent. They did this without the need to add additional staff. MoneyGram wanted to perform advanced analytics on financial transactions to trace patterns, and predict and prevent fraudulent activities. So they created a set of rules to filter by geography, and within high-volume transaction brackets to automatically flag transactions. This increased their ability to identify an interrupt potentially fraudulent transactions by 40 percent, and stopped more than $37 million in fraud transactions to date. They also realized a 72 percent reduction in customer fraud complaints in one year, and these new patterns when detected can be applied as a filter rule within hours or minutes. Another major financial firm, a trading firm, I can't mention by name. Had the opportunity to create huge price arbitrage in securities lending, if real-time price data was made available to traders. So they combined 18 different data sources to quote best price to brokers from other banks looking to cover short sales. Traders previously were dependent on weekly reports that took one analyst ten hours to build and led to stale pricing information. This solution, added $2 million a year in additional revenue from instant access to pricing data. A multi-billion dollar residential mortgage company called Primary Capital, needed instant access to mortgage portfolio risks to assess purchase decisions and sales performance. Their data includes information on one billion annual loans, with data spread in hundreds of tables across three different types of databases. Their analytics solution including natural language query capabilities, gives account execs real-time self-service visibility into the status of their accounts, reducing the time to track loan status from one to two hours, down to two to five minutes per day. They anticipate over 10 million in annual incremental revenue from the increased account visibility and productivity. Citibank needed to create a defensible revenue forecast model, to pass the comprehensive capital analysis and review the CCAR also known as the stress test. So they correlated and analyze 2,600 macro-economic variables, with revenue streams for dozens of business units, using Machine Intelligence software. The uncovered variable permutations that were hard to identify using incumbent analytics approaches, and shorten the variable selection process from three months down to two weeks. The results were that they were able to achieve the cleanest Federal Reserve test pass of all top US banks. The following day, Citi stock added nine billion dollars in market capitalization. They announced a dividend increase of 500 percent. Now Forbes wanted to expand the breadth and depth of market coverage for its readers with engaging digital content. Forbes receives earnings data on thousands of companies. So instead of having human writers write the reports, they used a natural language capability to automatically generate reports. This helped to increase site traffic, thereby broadening the audience and increasing ad revenues. They expanded this to cover over 600 stories each earnings season versus dozens they were previously able to write. This enables writers who spend more time on deeper analysis in investigative journalism. DBS Bank in Singapore needed a solution that can analyze the withdrawal data from 1,100 ATMs, to forecast activities and make better decisions. They created an analytic solution that integrates disparate operating concepts from manufacturing and logistics, as well as operations research techniques to optimize cache loading problems. Their ability to optimize the loading of ATMs, resulted in a 40 percent reduction in cash sent back to the bank, along with a 20 percent reduction in trips to replenish the network. Also, the number of cash-outs are down by more than 80 percent in which people can't get their cash because the machine is out of cash. This reduced customer wait time by 30,000 hours. American Express wanted to prevent corporate credit card customers from canceling, so they replaced static business rules, with more sophisticated predictive models. They were able to create 40 predictive models, analyzing 18 months of historical transactions and 115 variables including the customers industry, annual revenue, number of merchants paid, number of corporate credit cards, and charged points. As a result, they were able to identify 24 percent of Australian corporate accounts that would close within four months, and increase their ability to intervene to save the account.
