Manufacturers of consumer packaged goods (CPG) need to work closely with retailers of all sizes in order to get their products into the right hands, but that relationship is oftentimes complex and difficult to anticipate. Consumer demand can fluctuate wildly, causing unexpected swings in retailer demand and a host of confusions along the supply chain. Manufacturers either aren’t producing enough goods, or have a glut of product that unnecessarily fills up warehouse space. Neither are particularly good for business.
But the retailer-manufacturer connection has been getting better, mostly due to the influx of sales and purchasing data that major retailers like Wal-Mart and Target supply to their partners. This data can help manufacturers determine when — and in what amounts — retailers will make their next order, but forecast errors are still around 50 percent, and dealing with all that data isn’t getting any easier.
In order to make sense of this conflict, Manufacturing.net sat down with Rob Byrne, Founder and CEO of Terra Technology, to discuss the emergence of data-driven technologies that will help manufacturers be more agile in the face of a complex retailer landscape.
Manufacturing.net: Why is the CPG-retailer collaboration a difficult process?
Byrne: Your average grocery store has probably 200,000 or 250,000 products in it, and a few hundred manufacturers. The sheer scale of what’s going on here is difficult. If you look at Apple iPhones — you have black or white with three different memory sizes — that’s a lot easier. There has been a lot of focus over the years to collaborate, but by and large they’ve all failed because of the amount of effort and head count that you have to assign to it to make any real progress. If you can really assign one planner to watch one grocery store in one market, you think, “Okay, how many grocery stores are there? How many markets are there? How many planners do I need?” That’s the biggest issue: the scalability of it.
The average forecast error in consumer packaged goods is about 50 percent. If you think you’re going to sell 10,000 boxes of cereal one week, half the time you’re going to sell 5,000 or 15,000, and half the time it’s outside of that range completely. It’s pretty astonishing. We started looking at ways to improve that. One of the things we hit upon is what the retailers are looking at when they order from you. They’re not looking at sales a year ago — they’re looking at the inventory they have in their warehouses and what’s selling in their stores. By collecting that data, as a manufacturer, you gain a much better idea of what the retailers are going to do.
M.net: Why haven’t retailers and manufacturers started collaborating already?
Byrne: I think it’s a couple reasons. Retailers operate with very thin margins, and they tend to under-invest in IT solutions. The idea they should spend money to buy servers to supply data to their suppliers is a “maybe.” Also, retailers [have been concerned] that manufacturers would use the data competitively, because a manufacturer might be getting data from multiple retailers. That’s been more prevalent in Europe thanNorth America. The other issue, to a certain extent, is the quality of the data.
M.net: How can manufacturers get their retail partners to offer this data?
Byrne: We’re still in the early days. Wal-Mart is doing it, Target is doing it, and people are still talking to Costco about retailer data. People have started with their largest customers, and most manufacturers we work with have about 65 percent coverage for their sales. You still can’t get convenience stores or gas stations. I don’t think there has been a particularly good pitch historically — it’s been largely driven by customer teams who want access to the data. We’re really the first company to offer a structured approach — give us the data, let us crunch the numbers for you. Up until now it’s been a marketing tool (to determine) how a promotion did last month.
M.net: Are manufacturers taking these results to retailers in order to collaborate on new ways of doing business with each other?
Byrne: That’s starting to happen. Right now, if you look at supply chains, both the manufacturers and the retailers have lots of data. But they’re connected by really short-term, transactional pieces of information — I send you an order, you send me an advance shipping notice — there’s no logical planning layer that connects the supply chains. That’s what we’re offering. You want to extend your inventory optimization down to the retailer, you want to synchronize your demand signals at the retailer level and you want to optimize the orders going toward the retailer. [Our process is] a bit like a much more elegant, automated, granular version of vendor managed inventory.
M.net: How does using Terra help a manufacturer predict or anticipate when they need to ship and how much to produce?
Byrne: That’s the demand sensing — we’re using all that data from the retailer to predict when they’re going to order next and how much they’re going to order after that. By reducing that volatility by 30 to 40 percent, we allow the manufacturer to reduce inventory on hand because demand is more predictable. If your error for next week is 20 percent instead of 50 percent, you just don’t need to carry as much inventory. As you move down the supply chain, closer to the store, you really reduce what’s called the “bull whip” effect. By modeling the supply chain further down, you can reduce the volatility all the way back, so you really understand what’s happening at the end of the supply chain. That allows the manufacturer to respond much more effectively.
M.net: What kind of return are you achieving?
Byrne: We’re generally reducing the volatility by 30 to 40 percent. I think the key thing is that everyone is talking about collaboration, but you have to do it without people. You need a structured approach to dealing with the quantity of data, and you need to do it quickly.
After the interview, Byrne supplied Manufacturing.net with examples of how this kind of data-driven collaboration has affected some large-scale CPG manufacturers.
Unilever: In a presentation given by Unilever Chief Supply Chain Officer, Pier Luigi Sigismondi at the 2011 Gartner European Supply Chain Executive Summit in London, he noted Unilever North America removed 5 days of inventory using Terra Demand Sensing.
Campbell Soup: Campbell Soup’s implementation of Terra Demand Sensing and Inventory Optimization solutions resulted in a $20m savings in inventory in the first year.
Interview by Joel Hans, Editor, Manufacturing.net