Business Intelligence (BI) continues to redefine the tools of engagement across the retailing landscape. In the early days the value of BI was limited to historical perspectives. For example: An executive, needing to make a game-changing decision, turned to an analyst for help in defining key performance metrics and related variables. In turn, the analyst worked directly with IT to extract and transfer the requisite dataset; then load it into a behemoth computer for in-depth analysis. Days, sometimes weeks or even months later, reports landed on the executive’s desk providing insights into what was.
While early BI did provide value, its use was limited due to the processing power required for managing and analyzing such massive datasets. However, as the cost of computing declined, processing power increased, making BI and robust analytical applications readily available to a broader audience. Consequently, retailers and CPG manufacturers began extracting unprecedented insights from their information assets. Shopper and loyalty data, lying dormant in network hard-drives, was suddenly brought to life in the forms of consumer decision trees and behavioral segmentation. New data models began surfacing that combined heuristics with predictive analytics to anticipate - even optimize - future events. The world of insight was leapfrogging as the intelligence of “what-was” yielded to the wisdom of “what-will-be.”
Business Wisdom is Intelligence Applied
Using advanced analytics to extract actionable insights from big data can be found across the consumer packaged goods value chain. While the list of applications continues to expand, Willard Bishop considers the following to be the fastest growing segments:
1. Demand-Driven Pricing
2. Multi-Attribute Segmentation
3. Behavior/Demand-Driven Store Clustering
4. Shopper-Centered Merchandising
5. Customer-Driven Assortment Scaling
6. Macro Space Optimization
7. Customer-Driven Ad and Promotion Effectiveness
8. Predictive Web Analytics and Targeting
9. Social Monitoring and Sentiment Analysis
10. Dynamic Store Coverage
Most price optimization tools today are very good at leveraging elasticity science to dial prices up and down over time to maximize profitability. But, they often lack strategy, which is critical to create and protect price image. Without strategy and important guardrails, category and item pricing can get out-of-alignment over time.
A new approach has emerged that leverages demand forecast data and pricing strategy guidelines/rules to determine prices. This approach ensures prices fit in with the overall store strategy and it allows retailers to leverage advanced forecasting data to understand how changes in will impact demand.
Demand-Driven Pricing requires regular and promoted shelf prices for every SKU, as well as, demand data over a 52-104 week period. It is not for the faint-hearted but it will produce a stronger price image, greater foot traffic, and a process to maintain the necessary checks and balances that keep stores on the right course.
Single segmentation frameworks are making way for more complex multi-customer segmentations. Multi-customer attribute schemes are being used to develop strategic segmentations across a host of applications.
Retailer’s now leverage loyalty data to develop advanced customer-driven price segmentation strategies. These advanced analytics allow retailers to quickly learn who the ultra-price sensitive shoppers are, along with the most price insensitive shopper groups. Retailers know who to target, what items should be targeted, and what price points they want to communicate. It’s not the usual KVIs (known-value items) that make the list for price reductions. And, it’s not the typical items that make the list for price increases. The result is changing price perceptions, increased trips, and greater basket volume.
A number of retailers are also digging deep into understanding trips and trip missions. These trip mission analyses can be built using primarily T-Log data, but are turbo-charged with customer data. Retailers identify what trip types they are winning and losing. They can assess the profitability implications of wins and losses by chain, cluster, division, and store. Then, they make an investment decision on which trips should be targeted for growth.
Retailers are improving their investment productivity by developing customer-driven promotion segmentation plans. They classify who is promotionally promiscuous and who is not. They are able to assess the impact of promotions, not just lift, but the breadth of shopping in the current basket and the potential impact on baskets over time. They are identifying the cherry pickers (items only bought on deal) and eliminating promotions that are not contributing to the bottom line.
Behavior/Demand-Driven Store Clustering
Varying assortment and space based on store size is no longer an acceptable practice. The need to personalize and localize assortment now requires retailers to truly understand store-by-store purchase behavior.
Retailers are building as many as seven store clusters. They are creating category-by-category assortments for each cluster, powered by chain-specific consumer decision trees.
More than two-thirds of grocery categories have a fundamentally different buying pattern across stores. Using an average set means more than 50% of stores are under- or over-merchandised. By adjusting assortment and sets to specific clusters and buying habits, retailers are able to reduce out-of-stocks by 10%-30%, drive sales 2%-4%, and make dramatic improvements in inventory reduction.
This advanced analytical capability requires good demand data and card data to track loyalty and brand switching.
Retailers with T-Log or loyalty card data are sitting on a gold mine that can improve the way the store is merchandised. Most retailers use this data today to understand purchase behaviors and target (one-to-one) promotions that create greater store loyalty. This big data set can also be used to better understand how consumers shop the store, leading to new insights that produce more shopper-friendly sets and improved store layouts. New processes and “pattern recognition” tools are emerging to tap this data and create optimal flows and adjacencies based on trip dynamics, basket profiles, and other advanced analytics. This facilitates new merchandising insights across, and within, different store cluster groups.
Shopper-Centered Merchandising requires detailed transaction data or loyalty card data by store cluster, store, and by week to identify trip-demand patterns.
Customer-Driven Assortment Scaling
Customer-driven assortment scaling has been around for a while and it’s the old dog in this group of leading-edge analytics. However, it still is an area that retailers are trying to advance. The goal is to create store-specific planograms that better match the demand patterns of the individual store and surrounding customer base. This tool creates more localized and personalized shelf schematics, which lead to improved shopper satisfaction, increased purchases, and improved loyalty.
Emerging algorithms are also improving performance by reducing the “misfires” and “reworks” created by the automation; however, the need to tailor assortment with minimal human intervention remains.
Adopting Customer-Driven Assortment Scaling requires store demand data (weekly, daily, baseline, incremental), POG data, and access to good market data.
Macro Space Optimization
Our industry spends a great deal of time building and maintaining planograms. Even with all of this work, most organizations really can’t answer the question: Is that set optimal for this store and if I reduced space or grew space, what is the best source for the space change? Macro space optimization looks at the trade-off potential across the entire store or department and produces recommended set sizes for every category to optimize the available space. It accounts for different cube, inventory requirements, demand patterns, merchandising rules, growth, etc.
Recent advances in macro space have produced more accurate space recommendations. New tools take into account assortment and segmentation data that ensures retailers have the right amount of coverage across the right product segments. Some of the early tools simply cut off the tail of the curve and did not adequately protect variety.
This advanced analytical tool requires weekly-demand data for every SKU across the store. It also requires clean store maps and accurate POG files with good dimension data, cubes, case packs, growth rates, etc.
Customer-Driven Ad and Promotion Effectiveness
Get a group of retailers in a room and ask them what their objectives are for the front page ad and you get a wide variety of answers. With the advent of advanced analytics, some retailers are using smart metrics rather than settle for the typical “this year” versus “last year” or “this week” versus “last week” tactic to set front-page ads. Retailers now understand which customer segments are buying which items, the breadth of store purchasing driven by the ad item, penetration, traffic growth, and profitability by segment.
These new analytics help optimize the items that appear in the ad, appear on end-aisle displays or side-stack displays. It has eliminated advertising and promotions that don’t generate traffic, incremental purchases, or additional trips.
This advanced analytics requires good loyalty card data and market basket analyses tied to weekly ads, features, and promotions.
Predictive Web Analytics and Targeting
Historically, targeting was done on a basis of category usage and loyalty level. “Loyalists” received one offer while “switchers” got another, usually a more generous offer. Switchers required a larger incentive to affect behavior. Competitive brand users were given even more money to switch. The reality is, as much as 30% to 50% of a brand’s heavy users actually turnover during a year. It’s much more important to understand the fundamental dynamics of these core users.
Retailers are now leveraging predictive analytics to drive targeting. They’re making much more informed decisions on who will respond to a variety of offers, their likely actions, and by which channel consumers prefer the communications. All of this is done in a multi-channel context. So, it’s not just behavior-based analytics, but a whole new school of web analytics that drive ad and offer placement.
As food grocers continue to develop their CRM and digital experience, the science from these other verticals will help them improve target selection, increase redemption rates, reduce campaign costs, and drive far more effective long-term engagement.
Social Monitoring and Sentiment Analysis
Social monitoring provides a totally new way to evaluate how customers talk about retailer stores and their brands. It provides both good and bad news. This “smart listening,” provides retailers early warning signals on potential problems and an ability to react more quickly. It also provides a window into new product ideas from their most valuable customers and previously unknown brand advocates who are blogging about their brand in more positive ways than they could ever imagine.
As the digital experience penetrates more food-based retailers, their ability to listen to this “noise” will increase dramatically. New web scraping technologies provide an efficient and effective way to monitor this social chatter.
Dynamic Store Coverage
This store ops capability is allowing supplier partners to better align resources and in-store labor against the stores, categories, and activities that drive the greatest value. Historically, we have allocated merchandiser time on a rather static basis. Each store gets an hour and we cover the same activities across all categories in that hour.
iPads, GPS, and other demand information are being leveraged to apply the right amount of labor and the right activities in the right categories and at the right stores.
In the near future, suppliers will create dynamic merchandising routes like they create dynamic transportation routes today. The benefit will be shifting resources where the greater demand and needs exist.
No company is applying all of the advanced analytics/BI capabilities described above; however, many service providers are now offering their applications via Software-as-a-Service (SaaS) in order to streamline adoption and reduce entry costs. This, combined with cheaper data storage options such as data marts and data warehouses, will continue to expand the use of decision support systems across the value chain.