Terra on a Tear


The holy grail of consumer packaged goods (CPG) companies (or any manufacturer, for that matter) is to become more responsive to customer demand while reducing inventory. Learn how new solutions such as Demand Sensing, or what we call near-term planning, now make that possible.


For decades, CPG companies created long-term forecasts, with levels of granularity in months and maybe weeks. Conquering the mountain of inventory and learning to be responsive have always been challenging for CPG companies. Demand variability, in practice, is usually addressed with inventory. There was little opportunity to be more responsive and also keep inventories low. Though progress has been made over the years, inventory levels for most CPG products can be from 1 to 6 months!

It’s a bit of a myth, though, that inventory equates to customer service, since the question always arises, exactly which product does the customer demand? It may be the one you don’t have in stock. In addition, this is a costly approach. Months of inventory, stuffed into warehouses, impact margin, and can be very inconvenient with perishable or end-of-life goods. End-of-life, or after-sell-date inventory, is not visible, missing that last chance to earn revenue, and may ultimately be trashed, which can cost millions.

But no more! Terra Technology has been on a tear, grinding down the level of inventory for some of the most admired brands in the world. We recently talked with the company and learned how they have evolved. Some exciting news has come out of this Connecticut-based company, including a European presence to support their global CPG client base.

So what is Terra doing that traditional forecasting software firms are not?

Traditional forecasting packages and approaches were just not architected to sense this ‘near term’ demand, sorting through the layers of data, crossed signals and huge volumes of POS. Nor were they designed to determine whether there is existing inventory, in the plethora of warehouses, which can be used to economically fulfill that demand–now.

So, why don’t traditional forecasting solutions deliver this? For a number of reasons. Firstly forecasting packages are not designed to be “responsive.” That is, they are designed to create a forecast. They do not then analyze short term demand (changes in forecast, new orders, nuances in demand) and based on inventory positions (forecasting packages are not actually looking at real inventory) create a new plan and execute it. In other words, forecasting systems are not designed as execution systems.

Terra Technology has dedicated itself to addressing this very type of challenge. And, as it turns out, that is a very good place to be. The company has grown, even in this down economy, by getting it right, and adding value to even some of the most supply chain sophisticated and progressive companies out there.

P&G is one case in point. Without sharing too much of P&G’s data, they credit their use of Terra’s Demand Sensing solution for helping to reduce forecast errors by around 40%! And the subsequent inventory saving of ~10%.

Demand is a very hard number to derive when you are selling into a rich network of channel partners and retailers. Many organizations interpret withdrawal of inventory from a warehouse as a signal to replenish, when it may actually be just moving inventory to the back of a store. In addition, POS data can be inaccurate, late, and so prolific that it is difficult to interpret. So interpreting these factors and translating them into what is a real demand signal is the work. Figure 2 is a simplified version of the span of information and the audience for the solution. In other words, traditional forecasting systems are not designed to sense demand.

Terra Graphic
Figure 1: Multi-echelon View of Demand and Supply

Sitting between various data sources–warehouse inventory, in-store inventories, plant inventory and other inventory stores, the demand sensing engine can determine demand and update the forecast daily based on meaningful events. Then everyone executes to the most current, accurate numbers, not to something agreed to a month before. This requires new planning system algorithms that take into account not only seasonal trends from previous years, but also orders received yesterday. There are a lot of nuances in interpreting demand, and a system has to learn them. In addition, with a more accurate product fulfillment schedule, this can be used to improve transportation schedules as well.

The users of such data are the manufacturer, the wholesaler and probably the retailer itself. Although it may be purchased by the brand company, everybody wants that great data!

To view other articles from this issue of the brief, click here

Scroll to Top