The on-shelf availability challenge
At the heart of retail shelf availability are two core processes: forecasting consumer demand and aligning replenishment systems with these forecasts. The first still tends to operate through traditional and simple historical forecast models. The latter has some hidden problems that cause nightmares to store managers. Both are undergoing a change in the world of retail
A customer walking down an aisle in search of a product seems like a mundane activity. But, in reality, it is a “moment of truth”. Maybe she finds what she is looking for and puts it into her basket. Or, maybe she doesn’t. Does she then defer her purchase, choose a substitute or just go somewhere else? This is the worry of the retail industry. While the problem of keeping a shelf stocked seems a simple one to solve, in reality it is anything but. At the heart of retail shelf availability are two core processes: forecasting consumer demand and aligning replenishment systems with these forecasts. The first still tends to operate through traditional and simple historical forecast models. The latter has some quiet hidden problems that cause nightmares to store managers. Both are undergoing quite a change in the world of retail.
Why forecasting is usually off the mark
Forecasting is easy to do if things do not change. If the same customers walked in on the same days and bought the same items, life would be easier. Boring, perhaps, and shareholders might be unhappy, but it would definitely be easier to manage operations. In a slowly changing world, forecasts based on traditional time series data or thumb rules are adequate.
Complications enter the picture because of growth, promotion and innovation. As businesses want growth, we want our existing customers to buy more in our stores and for new customers to start coming in. The problem with new customers is that we do not know their buying patterns in advance. Even existing customers have changing wants and needs. This unexpected demand pattern can lead to items going out of stock.
Supplier brands run promotions and so too do retail stores. This can lead to unexpected buying behaviour of customers. Sometimes they buy more. Sometimes they do not. Again, it is easy for inventories to go out of whack.
Finally, innovation is the mantra for success in today’s world. Suppliers innovate with new brands, brand extensions and brand packs. Retailers, in turn, innovate with free gifts, loyalty schemes, price offs and bundling – all of which adds more uncertainty when it comes to predicting what will sell and planning inventories to ensure availability. When there is so much change in the system, it is hard to survive only on forecasts based on historical trends. Fortunately, today we have methods and tools to help do just that.
An example of looking outside & forecasting
In a recent exercise with an online retailer, we had built a forecasting model based on a combination of inputs – historical trends (yes, that was certainly there), but also the nature of competition in the different categories and customer search trends online. The model was tested against past data, using six months data to forecast the seventh month. When compared with the actual performance, the new forecasting model would have resulted in a near 9% increase in sales.
As the pace of change increases in the retail industry, be it product launches, variant launches or promotions, this kind of forward and outward looking techniques will become more essential to make accurate forecasts. There will also be issues around new products, where there is no historical data to use. In this case, the exercise gets more complex.
Where new products are concerned, future demand is dependent not only on consumer past behaviour but also on the actions taken to change that behaviour. Consumers who were using a product earlier will be induced to try out a new product. This will be through active interventions, such as media support, sampling, launch market selection and the positioning and price-performance relative to competition. All these factors will require different kinds of inputs for better quality of forecasting. New product forecasting can be complex. New category forecasting can be near impossible.
What happens once the forecast is made?
However, it is done, once a forecast that has been made, we need to get into the nuts and bolts of the system to ensure on shelf availability. This is where replenishment steps in. At its core, a replenishment system is very simple. When the stock level falls below a minimum threshold, it triggers a new order. The new stock comes in time to ensure there is no stock out. The theory is simple, but in reality, it takes a lot of discipline and considerable smarts to make it work.
The discipline comes in the master data and inventory management. This is about the accuracy of the records and the accuracy of the physical inventory. Firstly, is what is on the books really there on the shelf? Second, is it where it really should be? This is hard to ensure with the number of SKUs that retailers deal with. Adding to the confusion are special offers, displays and promotional activities. It is not surprising that around 75% of lowered on shelf availability originates from the retail store and not from upstream supply.
The smarts is about setting the threshold level. This seemingly simple task becomes complex because of scale and changes in demand patterns. The scale problem is easy to understand – when there are 100 SKUs someone can review numbers and set formulae for calculating stock levels. When there are 100,000 SKUs, it becomes more difficult. Are the masters accurate? How frequently should the calculation be done? Who will crosscheck the outcomes? These are difficult to do manually.
It is further complicated by demand patterns that change. If a promotion is planned and a promotional SKU is to be introduced by a brand, how is the regular stock phased out and the promotional stock phased in? Later, once the promotion is over, how is the regular stock reintroduced? When there is a festival season and a surge in demand anticipated, how should stock norms be reset? Similarly, if weather conditions become abnormal, do we scale down on some items? If stock norms were not actively managed, inventories would be constantly out of sync with demand. So when there are changing demand patterns, it is necessary to actively and dynamically manage stocking levels.
Fortunately, there are increasingly sophisticated tools to do such activities and these can be run off existing ERP systems. This provides the capability to tune the inventory levels closer to both long-term demand changes as well as short-term fluctuations that come up.
Why forecast when there is replenishment?
Several years ago, a senior marketing professional made a statement – “Since we run on a replenishment model, why do I need to forecast?” This doubt lingers in the minds of many people. There are two reasons for this. Firstly, as we have just seen, the faster the pace of change the greater the need for a forecast. Secondly, even in situations of relative stability, but also one of growth, it is necessary to forecast in order to address long lead-time items. This could be staffing levels, addition of new suppliers, infrastructure or logistics capacity. Simply having a robust replenishment method will not allow us to escape the need for more accurate forecasts.
In the 50s, the RAND Corporation introduced Material Requirement Planning (MRP) to manufacturing. It was a radically new approach to planning materials. The original theory was known for a long time, but the capability to calculate it did not exist in the past. So, when the early computers moved into the world of business, MRP took root and is a fundamental tool today the world over. Similarly, it is likely that the sun is rising on a new set of tools for inventory management and shelf availability in the world of Indian retail.
The author is Partner & Managing Director – Indian Operations, CGN & Associates India Pvt Ltd