Dr Pepper Snapple Group looks to automate category management
Through a special arrangement, presented here for discussion is a summary of a current article from the monthly e-zine, CPGmatters.
For the past six years, Dr Pepper Snapple Group has relied on a seven-stage category management process for space and assortments in stores. Now the beverage company wants to speed up the process. The goal is to continuously update the planograms by feeding more data into the system automatically.
"We want to make category management a continuous process, not a project-based process," said John Williams, director of category management for the company, recently at the Category Management Conference hosted by the Category Management Association (CMA) in Orlando. "We need to be more automated and then focus on insights and the shopper. That’s where we see it going."
Continuous Category Management is generally defined as a process that employs real-time information and instant alerts to enable trading partners to make informed category and product merchandising decisions more rapidly.
New business processes and technologies have enabled the automation of store-specific planograms in seconds rather than months. Meanwhile, data collection technologies from image recognition powered by on-demand robotic technologies that maps whole stores down to shelf positions can provide deeper insights into assortment management as well as optimally link the shelf to the supply chain.
Danny Halim, JDA Software’s VP of industry strategies, who spoke with Mr. Williams, said that while cameras in stores can monitor shopper behavior, they can also be used to view shelf conditions.
"The interesting thing about this is thinking beyond a snapshot of the shelf," he said. "What if we can get real-time information from all the stores? With a great deal of accuracy, we can understand the condition of the shelves: the compliance of the planograms on the shelves, the in-stock and out-of- stock conditions on the shelves. I think that is going to be the game changer."
The cloud could be another area of development for automation. Mr. Williams said, "As you know, an IT project can take a long time. Sometimes we might not have the resources. But put it into the cloud, and it becomes very automated. That’s where we see it going."
Continuous category management is another level to reach for; that is, having data feed into the process and allowing planograms to be created and reviewed automatically.
"We can also have analytics based on trends," Mr. Williams said, "not on what previously has happened. Days of supply: We could use a methodology that would be part of it. You would have the right days for a bottler delivery to the stores so you’re not out of stocks. We then take it from category to aisle and to the rest of the store."
Discussion Questions
In what ways may new data collection technologies reshape the future of category management? What parts of the category management process will benefit most from such real-time analysis and automation?
I think it’s great that technology we built over 20 years ago (at The Retail Alliance) is finally getting used. And that a speech I gave at ESOMAR in 2007 on automation and a focus on insights is finally getting adopted. But let’s be honest—nobody is going to change shelf sets on the fly on a continuous basis because labor costs would eat up any potential increase in sales. Once you’ve improved the shelf assortment and shelf layout for a retailer, the incremental gains you’ll see by tweaking it are very small.
There is a huge difference in having real time data and the ability to execute.
The definition provided by Karolefski was, “real-time information and instant alerts to enable trading partners to make informed category … [decisions].”
The operative word here is PARTNERS. Category management requires a partnership between retailers and suppliers.
While there is no question that Dr. Pepper Snapple Group and other CPG suppliers can significantly improve real-time data, do the retailers have the capacity, resources and systems to execute in “real-time” based on data?
There will be a law of diminishing returns for results realized from automated real-time data based upon the retailers’ ability to execute.
This is where technology is far outracing execution. It is nice to come up with new planograms daily, weekly, monthly. They can be sent to the stores to implement.
But who implements them at the store level? It is still a matter of arms and legs.
The challenge with retail automation is that the human beings in the process will always be the constraining factor. People with never move as quickly as bits and bytes. Just because it can be done does not necessarily mean it should. This is particularly true with shelf allocations.
I agree that technology which provides localized planograms and automated compliance feedback are important new capabilities. I disagree that frequent changes to planograms are a help to anyone. They mean additional labor for implementation, changes to shelf labels for unit pricing and increased confusion for night crews. For DSD items the factors are a little different but frequent changes in planograms still don’t seem like a good idea.
Like most generalities there must be exceptions. Some categories (sodas and drinks are one of these) have periodic promotion schedules that have a significant impact on individual item demand and require flexible shelf allocation. DSD items (again big in sodas and drinks) may be more easily implemented because they are being replenished by the vendor rep.
So my fundamental thought is that how you use these new capabilities will vary by category with most emphasis on localization and feedback, while implementing changes differently for each category.
These new data sources, such as high-resolution shelf images, can be very powerful. They will be especially powerful if integrated with accurate real-time demand modeling and inventory data. The combination of current shelf position, forecasted near-term sales, and in-store inventory can drive a proactive system for reducing out-of-stocks. But while inexpensive high-resolution imaging may be the newest piece of the solution, we should not underestimate the need for accurate real-time forecasting, and the intelligent algorithms that synthesize the information into a shelf-stocking and ordering plan.
Any technology that provides quicker and more accurate information is always to be desired. However, as several have already commented, it still comes down to execution at the store level.
There have been many developments that have provided this level of information (RFID, electronic pushers, etc.). However these tools, while providing instant (or near-instant) information, haven’t solved the issue of how to execute changes at the store level to improve sales.
I can see this helping many retailers that use daily shipments or DSD suppliers to stock their shelves. Having worked at a retailer that did daily shipments (urban market), we found that having accurate statistics helped improve out of stocks, shrinkage and spoilage.
Even if execution of changes can’t take place at the store level quickly, this technology will still benefit category managers to see how customers make their selections and to be able to come up with better long term strategies.
DPSG has always been scrappy and resourceful about their use of data in ways that extend and leverage their strengths. Rather than try to outspend Coca-Cola and Pepsi or purchase data for data’s sake, their legacy (going back to Rob Colarossi and Craig Hodnett) has been on using data in meaningful ways. I think that the use of data to better understand the SHOPPER will be the holy grail, far more than applications focused on space, pricing, promotion and product supply. While there are gains to be had there, it is my suspicion that they will pale in comparison to the gains to be had in truly getting a better handle on what motivates, incents and contributes to purchase decisions, pre-during-post shopping. Often easier said than done—however, that is more likely the bigger pot of gold.