This is a two-part series on vertical business intelligence (BI) trends
- Trends and tips for using business intelligence and analytics in retail
- Trends and tips for using business intelligence in financial services
A majority of multichannel retail environments are structured as separate profit-and-loss centers – online store; physical retail store; catalog sales; and not-for-profit, business or government-only sales. In fact, each typically has a separate revenue target, IT infrastructure and business unit leader. Moreover, those business leaders usually have incentive plans based on the revenue targets set for their individual units.
Compounding this conundrum, other departments within a retail environment – such as merchandising, marketing, corporate planning and supply chain operations – are typically shared services.
What's more, compared to other industries, retail is highly focused on inventory replenishment. Multiple sales channels require that retailers fastidiously avoid "stock-outs" – or as commonly viewed in online shopping carts, "out-of-stock" (OOS) situations – in order to maintain brand primacy and customer loyalty and steer clear of lost sales. For example, if a customer prefers Band-Aid brand adhesive bandages but a drugstore has in stock only NexCare, a 3M brand, the probability of the customer driving to another store is slim – especially when gasoline costs nearly $3 per gallon, or more.
Finding success with business intelligence in retail
The diversity of stock-keeping units (SKUs) makes retail merchandising a rich lode for business intelligence (BI) utilization. Retailers routinely parse sales data in an effort to better determine what will sell in the future. For instance, does a particular color or size sell more than another does? Do different colors sell better in one season or region than others? Does one store in a particular region do better than the others there?
When looking at sales, retailers frequently adjust the prism through which they measure success. As the consumer goods and retail sectors continue to expand their original business model from that of a single brand and/or brick-and-mortar store to include catalog and online channels, other challenges surface. Retailers as well as consumer goods brands have found that measuring success effectively requires tracking performance metrics for each discrete product or SKU through each distribution channel – in addition to a corporate level roll-up.
Finding outliers of success (or disappointment) can help identify best practices and highlight flaws. Always, the question of profitability looms, whether by customer or by product or by marketing campaign. For example, how much profit margin is realized per SKU for a certain in-store promotion? Are particular customers more profitable than others? Among other things, using BI allows retailers to improve existing affinity programs. They also can get answers to questions such as whether rebates or trade promotions on a particular item prompt customers to spend more during a particular shopping trip, or at a specific time of year.
Using business intelligence to optimize the supply chain
Hypatia Research has found that retailers often use both enterprise BI tools and Software as a Service (SaaS) applications to measure the effectiveness of marketing and promotional campaigns. Car Toys Inc. is one such example. Based in Seattle, Car Toys is a $150 million retailer of high-end mobile electronics, including stereo equipment, alarms, DVD players, GPS navigation devices, satellite radios and cell phones. Privately held, it employs more than 1,000 people in 49 locations in Washington, Oregon, Colorado and Texas.
The challenge. Car Toys tracks more than 100,000 SKUs. It has an ERP transaction system for supply chain and inventory management and uses BI software to track sales, inventory and marketing promotions. The company needs to ensure that its salespeople are cross-selling properly (that is, when they sell someone a cell phone, they also sell an earphone or a carrying case). Car Toys also compares foot traffic (which is tracked by door sensors) to sales at individual stores, as well as the results of regional promotions and year-over-year sales.
Operational approach. Car Toys had a Linux-based server running BI software, the only use of the open-source operating system in its offices. When it came time to upgrade the server, the vendor suggested that the company switch to a SaaS-based setup. Car Toys transitioned to a hosted system that it pays for based on volume of data and number of users, and it made the SaaS BI technology broadly available to employees – from the CEO down to the salespeople in each store. Out of 1,000 total employees, 500 use the SaaS BI tools. To run more complex metrics, the retailer engages with the vendor's professional services team.
Results. As a result of the changes, Car Toys is now able to extract information to deliver financial data to employees without paying for additional ERP licenses. And employees can take advantage of multiple views and reports. For example, a regional manager in northern Texas can create a view that represents only his stores, save it as an operational report and run it on a regular basis. Salespeople, meanwhile, can track how they're performing compared with their peers.
Other benefits include increased flexibility in defining and building specific operational reports and dashboards, and the ability to compare on-hand inventory vs. sales rates in order to predict optimal ordering estimates by SKU and location on a daily basis for critical decision making on product replenishment.
Using advanced analytics gives retailers an edge
Retailers strive to track and analyze "market baskets" and/or "attachment rates." Knowing the frequency of customer purchases and the store location, total dollar value and assortment of products bought in each store trip or online visit is of great interest for obvious reasons, including the following:
- Merchandising executives consider market-basket analysis – looking for relationships between the products that a customer buys at the same time – to be a key performance metric and use it extensively in planning store layouts, advertising and trade promotion campaigns.
- Attachment rates, which measure purchases of accessories and other supporting goods , are tracked in order to understand customer buying patterns by lead products.
- Measuring product correlation rates facilitates product sourcing, pricing and promotional decisions.
Research examples of attachment-analysis hypotheses used by retailers include the following:
- Customers who buy high-definition television sets also order installation services and/or longer warranties than those purchasing standard TVs.
- Women always purchase matching shoes when they buy dresses priced at more than $250.
- More often than not, customers who buy computer equipment also select peripheral items such as a mouse, storage devices and printer ink at the same time.
Companies contemplating investments in BI and analytical technologies that will be used to leverage multichannel information should carefully consider the following recommendations:
Define and standardize performance metrics on a corporate level. Corporate-wide agreement on how metrics are defined and calculated is as necessary as enterprise standardization of data dimensions across data marts. Failure to achieve this renders any analysis and insight derived from BI applications and reports not credible.
In a prime example taken from our research, activity-based costing was the preferred methodology for calculating product category profitability at one retailer. However, Hypatia found that one team included the actual cost of both production and raw materials in its calculation, while others added in an average cost of sales per product SKU. Cross-functional dysfunction resulted.
In another case, an office supply retailer utilized specific marketing metrics to calculate "uplift" from a trade promotion run on printer ink cartridges and internally announced the promotion as a rousing success. However, at the profitability level, the company actually lost revenue on the promotion as subsequent market-basket analysis revealed that the promotion didn't influence shoppers to buy other products with higher profit margins. In short, the retailer paid for expensive media advertising, store signage and direct marketing and reduced the price of the printer ink below cost – and realized both a negative return on its investment in the promotion and a revenue reduction.
Establish processes for updating, synchronizing, cleansing and normalizing all types of information. That includes data about your customers, products and supply chain partners. Gleaning market intelligence at a granular level, derived via advanced analytics, data mining and operational reports, requires that retailers have processes in place to track and store accurate, updated and normalized data prior to analysis.
Ensure that your organization takes out the trash. Boosting decision-support accuracy and credibility means ending the dissemination of bad data throughout a multichannel environment. Remember, it's not about data garbage in, data garbage out – it's about information garbage in, information garbage everywhere.
About the author: Leslie Ament, co-founder of Hypatia Research LLC, is a customer intelligence management thought-leader and industry analyst who focuses on how organizations capture, manage, analyze and apply actionable customer insight to improve customer management techniques, reduce operating expenses and accelerate corporate growth. Her research coverage includes: business intelligence, media intelligence/search/text analytics, CRM, Web analytics, marketing automation and customer data management/data quality.
Previously, Ament served on management teams and led global marketing and market research groups at Demantra Inc. (acquired by Oracle), Arthur D. Little Management Consulting, Harte-Hanks, Banta Corp., International Thomson Publishing (Chapman & Hall, U.K.) and Carnegie Hall Inc. She is a member of the American Marketing Association, the Society for Competitive Intelligence Professionals, the Customer Relationship Management Association, the Arthur D. Little Alumni Association and the Software & Information Industry Association. She also is a DataShaping Certified Analytics Professional and a board member of the Product Management Association.
Ament completed her doctorate as a Phi Kappa Phi member at the University of Illinois at Urbana-Champaign. She received her master's and bachelor's degrees at Indiana University in Bloomington.
Scores of companies offer analytical tools, platforms and services. For information on vendor selection criteria, research products or scheduling an analyst briefing, contact Hypatia at ZGR@HypatiaResearch.com or Research@HypatiaResearch.com.
This was first published in October 2009