Today hedge funds use alternative data to shape their investments. Why can’t organizations use it to shape their own future?
In the fall of 2015, Chipotle (an American food chain famous for its massive burritos) was rocked by a series of health scares. From August to December, norovirus and salmonella broke out in 19 locations across California, Massachusetts and Minnesota. The catastrophe reached a peak in October, when reports of e. coli outbreaks started to trickle in from Chipotle locations across Washington and Oregon. After 6 outbreaks across 12 states, over 500 people sickened, and a CDC investigation, Chipotle’s reputation was in the toilet.
As the public panicked, Chipotle’s stock price fell off a cliff, dropping over 50% from July 2015 to January 2016. Same-store sales fell nearly 15%, costing Chipotle $72 million in revenue. After a clean bill of health from the CDC in February 2016, Chipotle responded with an aggressive marketing campaign, including full-page apology letters in newspapers and free burrito coupons. Would it pull customers back into stores and Chipotle’s stock price out of its freefall?
Traditional sources of data were at a loss to answer this question. Wall Street surveyed hundreds of people, but the responses showed mixed sentiments. No food chain had experienced so many outbreaks, across so many locations, in such a short time, so historical data was no help. Social media sentiment was clearly against Chipotle, but would people’s outrage fade the next time they were hit with a craving for a Chipotle burrito? After all, Jack in the Box’s stocks rebounded shortly after its 1993 e. coli outbreak (which sickened 600 people and led to 4 deaths).
This is when alternative data — a phrase that few people had even heard of — started to shine. Orbital Insight, which uses satellite data to monitor the parking lots of over 50 U.S. retailers, was one of the first organizations to notice a drop-off in visitors to Chipotle. Meanwhile, Foursquare used check-ins and background location data collected by its apps to forecast that sales would continue to drop in the first quarter of 2016. They were right — Chipotle’s stock continued to fall and still hasn’t recovered two years later.
Today, alternative data is mainly used by hedge funds to shape their investments. But why can’t organizations themselves use this data to shape their own future? As alternative data becomes more accessible, there’s no reason why this isn’t possible. Here are 6 leading organizations across the world are using alternative data to take control of their own future and stay ahead of their competitors.
Use case: Managing sales, stock, shipping, and more
The world’s largest retailer, Walmart has been investing heavily in both managing its own data and integrating alternative data to solve problems across the company. It processes over 2.5 petabytes of data every hour, one of the world’s largest collections of data. (For comparison, just 8 hours of Walmart’s data is larger than all the data in the Internet Archive Wayback Machine, which stores over 500 billion archived versions of millions of webpages since 1996.)
Walmart’s largest data initiative is the Data Café (which stands for “Collaborative Analytics Facilities for Enterprise”), an data analytics hub located in its Arkansas headquarters. It pulls data from over 200 internal and external sources, including weather data, economic data, Nielsen data, telecom data, social media data, competitor data, and local events databases.
The system gives automated alerts when metrics fall below set thresholds. Then the team can quickly figure out what caused the drop and how it can be fixed. For example, a grocery team wanted to know why sales of a particular product had suddenly declined. By looking into the data, they quickly realized that the product was accidentally listed at a higher price in some regions.
Walmart uses this massive data collection to solve problems across the company, such as inventory management. Alternative data is crucial for understanding consumer interest, which helps Walmart stock the right shelves with the right products at the right time. For example, they correctly predicted people’s interest in cake-pop makers by analyzing Facebook and Twitter messages. This information was sent to the buying team, who could immediately act on it.
Walmart also uses their collection of internal and alternative data to figure out how to reduce shipping costs and optimize shipping routes, which products should be discontinued, which brands they should carry, how to reduce the amount of time it takes to fill prescriptions, how to anticipate store traffic so registers are adequately staffed, and more.
Industry: Consumer goods
Use case: Advertising
According to Neudata, one of today’s alternative data trends is phone location tracking. A famous example of this in action is Colgate’s advertising at the Maha Kumbh Mela.
One of the biggest gatherings in the world, the Kumbh Mela is a mass pilgrimage where Hindus bathe in one of four holy rivers in India. The standard Kumbh Melas happen every 12 years, but the Maha (Great) Kumbh Mela only happens every 144 years.
Like numerous other companies, Colgate set up a booth at the Maha Kumbh Mela to advertise to these millions of pilgrims. The problem was getting people to the booth and keeping them there, a hard task given the sheer number of brands and booths competing for everyone’s attention.
To attract and engage with pilgrims, Colgate created a virtual geo-fence around the fair. Any Airtel phone that entered this 5 km area would automatically get a call. The call played a recorded message from radio personality Amin Sayani, which told people to go to the Colgate booth for free samples, entertainment, and prizes. Once Colgate started this campaign, foot traffic to their booth increased by 300%, and a total of over 700,000 people visited the booth.
Use case: Predicting shipping rates
RR Donnelley is a Fortune 500 communications company that provides marketing and business communications, commercial printing, and other communications services. But today it’s using alternative data in an unexpected arena — shipping.
Over its 150 years of operations, RR Donnelley often had to transport massive amounts of documents, marketing materials and other items for their clients. Over time, the company realized it was good at managing independent shipping, so it opened an entire logistics division. Now it ships everything from dog food to refrigerators for its partners.
In a world dominated by shipping giants like UPS and FedEx, RR Donnelley could only win shipping contracts if it provided great rates. The problem was that the division didn’t know how to set these rates. If it pitched cautious rates that were too high, it would lose a bid. If it pitched low rates, it risked losing money if costs turned out to be higher than expected.
To predict the perfect shipping rates, the company turned to alternative data and machine learning. It hired a team of experts, who set up a complex multivariate predictive model. The model used both historical company data (on prior shipments and pricing) and alternative data (on weather, market conditions, and fuel prices) to predict freight rates seven days in advance with a 99% accuracy rate.
According to Ken O’Brien, CIO of RR Donnelley, the project cost $200,000 to set up, but it paid for itself in just the first month. Now, RR Donnelley is winning 4% more bids and it expects to quadruple the size of its shipping business from $4 million to $16 million.
Industry: Video games
Use case: Financial prediction
For years, Electronic Arts (the American video game company behind games like FIFA, Battlefield, and The SIMs) was vilified by gamers. Bypassing regulars like AT&T and Walmart, EA was overwhelmingly voted the Worst Company in America in both 2012 and 2013. (In the final 2012 round, EA beat Bank of America with 64% of the vote.) This was the first time a company had won this “award” two years running.
While announcing EA’s second win, The Consumerist seemed awed by the sheer hatred for EA: “When we live in an era marked by massive oil spills, faulty foreclosures by bad banks, and rampant consolidation in the airline and telecom industry, what does it say about EA’s business practices that so many people have — for the second year in a row — come out to hand it the title of Worst Company in America?”
Why did people hate EA so much? One cause was enormous criticism of EA’s “mediocre products”. Critics and gamers had recently roasted multiple aspects of Mass Effect 3, Dead Space 3, and SimCity 5. With this history, it’s no wonder that EA would want to pay attention to its fans’ thoughts before they publicly vilified the company.
To anticipate gamers’ reactions, EA turned to alternative data. They worked with Irish research firm Eagle Alpha, which analyzed 7,416 comments on a Reddit gaming thread in October 2015. The firm focused on understanding fans’ reaction to the upcoming Star Wars Battlefront game, which was then in open beta. (With over 9.5 million players, this was EA’s largest open beta to date.)
Based on Reddit data, Eagle Alpha predicted that EA would surpass its sales projections for the game. In response, EA increased its projected revenue for the year, citing fans’ excitement. “Based on the ongoing strength of our business and reception of Star Wars Battlefront, we are raising our full-year outlook for the second time,” said EA’s Chief Financial Officer Blake Jorgensen.
Ministry of Petroleum & Natural Gas, Bharat Gas, Hindustan Petroleum, and Indian Oil
Industry: Oil and gas
Use case: Opening distribution centers
Nearly half of India’s households use a chulha (a traditional firewood stove) for everyday cooking, which is equivalent to smoking 400 cigarettes an hour. But when the only way for women to get clean cooking fuel is to walk to the nearest LPG (propane) gas center up to 20 kilometers (12 miles) away, it makes sense why they’d choose to cook with firewood.
To solve this problem, the Ministry of Petroleum and Natural Gas (Government of India) and India’s 3 oil companies decided to open 10,000 new LPG centers. They partnered with SocialCops to use data to figure out where to place these centers. The centers needed to be profitable, since they were operated by small entrepreneurs, but they also needed to be placed so that every household in India would be within 10 km (6 miles) of an center.
To figure out which centers would be profitable within a few years, they used the oil companies’ internal data about supply and demand, sales, and consumers. They then added alternative data to figure out which possible locations would reach the right people. The data came from 600 sources, covering affluence, infrastructure, roads, ATMs and more, plus 17,000 geotagged locations of the existing LPG centers.
After complex data merging, their algorithms identified the best locations for these 10,000 new LPG centers, which are currently being commissioned and opened.
Industry: Consumer goods
Use case: Managing sales, stock, and merchandising
One of the world’s largest consumer goods companies, Unilever is constantly looking to innovate on how it stocks and sells its over 400 brands. One of its subsidiaries, Hindustan Unilever, is using alternative data to take on this challenge.
Hindustan Unilever (HUL) started an analytics initiative called Project iQ to create perfect stores and help its sales force make shorter, more effective sales calls. Each of the 12,000 urban sales agents carries a tablet, which tells the agents what to do at each store they visit. It shows which products are likely to sell, and which projects will run out of stock and need restocking. It even gives each sales agent an optimized schedule of stores to visit each day.
HUL’s merchandising agents also carry tablets, which tell them how to create the “perfect store” by arranging items properly on store shelves. This is critical, since HUL found that a neat, segmented arrangement of similar products helps boost a store’s sales by 30%.
Project iQ’s store-level recommendations are backed by data from HUL’s one and a half million outlets, plus alternative data on consumer buying behavior, buying patterns, and neighborhood behavior. This data powers various in-house predictive models, such as the company’s Cross-Sell Recommendations (customized predictive models that indicator the selling propensity of a product) and Stock-Out Analysis (the prediction of stock-outs based on closing stock, selling rate, and other parameters).
Soon after its launch, Project iQ showed promising results. In the quarter that Project iQ launched, HUL’s profit after tax jumped 22% (compared to the same quarter in the previous year). The project won the 2011 Global Unilever IT Award for “releasing the power of information”. Investors have even taken notice, changing their investment advice. Brokerage Kotak changed its recommendation for HUL from “sell” to “add”, Angel Broking marks it as “accumulate”, and HDFC Securities advises “buy on dips”.