Now, let's turn our attention to the retail industry. The retail industry, of course, comprises many different types of businesses from grocery, to clothing and accessories, to department stores, online retail, discount retail, electronics, food, home goods, specialty, et cetera. The types of analysis that's done in retail is perhaps more varied than in almost any other industry. From market analysis, to pricing and promotions, to understanding customers, competitors, to determining product mix, to managing inventory, stock turnover, understanding and managing distribution channels, laying out and designing stores, improving customer service, doing location analysis, loyalty programs, merchandising, supplier management, and understanding global and local trends and preferences. Let's take a look at the shoe store, DSW. They wanted to get early detection of quality issues with products to increase customer satisfaction and reduce damages. So, they created an analysis of inventory, customer returns, and sales data at a granular level, by store, by product, by color, by weight, by packaging, et cetera. They used a pattern analysis technology to detect issues across all stores and send specific descriptive insights directly to relevant users, combining DSWs best practices on how to resolve it. As a result, they are able to determine damaged products much earlier and generate over a million dollars more and vendor-provided credits for damaged products. This reduction in faulty products also resulted in increased customer satisfaction and minimized damage to the brand's reputation. Some years ago, the large department store Macy's wanted to better react to changing market conditions with near-real-time product pricing. The data they had included sales and inventory data on 73 million items, growing at an annual rate of 50 percent. Their data science organization generated and tested hundreds of thousands of analytic models on granular data versus the dozens that they previously had implemented. The results were that they were able to reduce the time to price items from 27 hours down to about one hour. A $9 billion US convenience store wanted to develop ongoing forecasting processes by market product category and region to decrease inventory write-offs and out of stock situations. They analyzed four years of historical sales data in five product categories from 350 convenience stores. Using an advanced analytics solution from preliminary software, they were able to analyze over a million external datasets, plus historical data and weather data. They created several dozen predictive models within the first month. As a result, they were able to improve the monthly accuracy of their forecasts to 98 percent by identifying leading drivers of major inventory items, and they were able to determine the impact of weather and create what-if weather models. An Indian retailer called Croma was dealing with the problem of plenty. Amid varied product offerings, customers were confused and conversions had suffered. They deployed an analytic solution to help them understand the customer's social profile, brand affinity, recent activities, household, and macroeconomic data. For their product catalog, they utilized the mix of natural language processing, semantic technologies, and machine learning to match and enrich the product understanding. As a result, in six months' time, the revenue share was up nearly 25 percent and conversions shot up by a staggering 217 percent. This deployment also resulted in a 29 percent increase in orders based on automated recommendations. A European bakery called Vaasan needed to accurately forecast fluctuating sales orders to avoid out-of-stock or excess inventory situations. They use near-real-time and historical customer data to create rolling sales forecasts and identify patterns in demand so that the company could have the right inventory levels to address the fluctuating demand. This resulted in a 30 percent increase in sales order fulfillment, and also the reduction in response time. They're now able to achieve an on-time delivery target of 98.5 percent while reducing the risk of lost sales by aligning resources better. A $3 billion Russian electronic store also wanted to improve its forecasting. Their data included 1.5 million point-of-sale transactions per month, for 420 product groups and sales of 8,000 products from 400 stores. They used a predictive analytics technology from SAP called Infinite Insight, against all of the data that they had stored in their SAP system. They created 500 predictive models per month, resulting in a 10 percent improvement in forecast, accuracy, and improved inventory management, pricing, sales, and staffing. Even small grocers can get in on the action. A company called Iron Cactus wanted to improve in-store customer experience. So, they were clever in using historical video feeds from their existing security cameras and analyzed them using a product called Prism Skylab to understand shopper profiles such as sex and estimated age and shopping traffic patterns. The heat maps they created helped them identify customer wait times, enabling the business to improve store flow, by optimizing relative product placement. From this analysis, they were also able to improve employee assignments and scheduling. Walmart wanted to help their shoppers find what they're looking for more quickly. They integrated product and category popularity scores generated from social media feeds using text mining. This machine learning-based semantic search capability on clickstream data from 45 million online shoppers, enabled them to achieve a 10-15 percent increase in online shoppers actually completing a purchase or reducing shopping cart abandonment. Dollar General is one of several companies now actually monetizing its data by making it available to its CPG partners. They placed billions of rows of point-of-sale data, inventory, promotion, and other data into the cloud-based data store for partners to analyze in a common spreadsheet-like format. Their CPG partners now could subscribe to the data and help them better market and promote their products. Now, Dollar General has a self-funding analytic solution. A large home improvement store recently spoke to me about how to grow revenue through the improved targeting and cross-selling of yard care products. I recommended to them that they should layer their point-of-sale data and warranty customer data with map information, geographic information, to identify customers that have large foliage coverage and large driveways. This could help them identify new DIY or do-it-yourself customers to cross-sell, leaf and snow blowers or partners' products. It's yet to be seen how the solution will help them improve direct marketing campaign performance by focusing on customers with actual needs and specific behaviors. It can also help them develop high-value partnerships with their suppliers. We all may think that we have free will to select whatever items we want as we're driving through the drive-through lane of a fast-food restaurant, but perhaps that's not the case. Now, I can't mention which fast-food company this is, but they wanted to optimize customer and margin revenue. So they continually analyze video from the drive-through lanes and on the street to identify customer drive-by behavior and situations. They also outsource credit card data to analyze pre and post-visit customer behavior in aggregate de-identified. Most interestingly, they update the drive-through lane order boards as needed to feature more expedient items to make when the lines get long, and higher-margin items when the lines get short. They also dynamically update pricing.
