The Consumer and Demand-Driven Retailing

Much has been written about Demand-Driven retailing.  In this section, we delve into this and talk about how an accurate forecast that looks at a variety of influences on consumer demand is the start of the Demand-Driven Supply Network (DDSN). What are the elements required to achieve demand-driven retailing and how does the ClarityRetail engine help?

The Intersection of Demand Chain Management and Consumer-Centric Merchandising

While these are two distinct disciplines retailers need to possess, there are overlaps and cross-benefits that can occur when these functions are implemented concurrently. 

Let’s look at some basic questions first to get us thinking about demand-driven retailing and the ingredients needed to get there.

Are You Making the Most Effective Use of Your POS Data?

One of the most valuable sources of data exists in your daily POS (Point-of-Sale) movement.  While amongst obvious phrases, this might perhaps be the most obvious, it is surprising how many retailers today are not using the POS Sales data effectively.  Sure, we update accounting and ERP systems with it, but how is this data used by those systems to better manage and plan in a demand-driven increasingly consumer-centric world? 

An effective consumer-centric retailing framework starts with a good Demand Forecast.  How else can we begin to tackle the rest of the supply chain challenges faced by retailers today without an accurate picture of consumer demand at the moment of truth?  Your POS system represents that “moment of truth”, the precise time when a customer made a decision and executed on it by paying for the product.

How else can we start tackling the rest of the processes required such as assortment planning, pricing, promotion planning and space allocation when we lack the access to timely data required to do so?  Also, without an understanding of store-specific consumer demand, our best efforts will result in a theoretical model which we think should apply to a group of stores or all the stores in the chain.  As you begin to execute on this theoretical model, things begin to fall apart as stores have too much inventory of certain products and not enough of others.  Out-Of-Stocks (OOS) start to climb and inventory handling costs increase.  While there may be a certain amount of consistency in sales patterns from store to store, depending on the retail segment you are in, frequent divergences will occur that break the aggregated model you have constructed.  This is especially true in the grocery/supermarket segment but can also apply to clothing, sporting equipment and the like depending on the season and demographics present around a particular store. 

What does all this imply?  To get a handle on store-specific consumer demand, you need to analyze POS sales, price and promotions BY STORE.  This is exactly what ClarityRetail does for you each and every night.

Let’s Start with Baseline Forecasts

There has been much discussion as of late with respect to whether or not baseline forecasts offer any value to a retailer given the amount of promotional activity and how customers are behaving with respect to loyalty. We here at ClarityRetail still believe that there is a value to a baseline forecast but before we delve further into that, let's define what we consider a baseline forecast to be.

A baseline forecast assumes that the item in question is currently not on any form of promotion whether retailer-driven or manufacturer-driven.  This means there are no coupons for this item floating around in circulation, no current offers available in-store and the item is not featured currently in any weekly ads.  The sales volume generated during this period is what we would consider baseline data in that it is absent of any promotional influence.  The trick here is however that non-promoted movement must be clearly distinguishable from promoted movement within the data.

Of course we all understand that there may still be other factors at work that will influence demand for this particular item. For example there could be promotions taking place for competitive products within the same category or a competing retailer may be running a promotion on the exact same item or one that is very similar. Clearly either of these two scenarios will affect sales volume of this item so the argument put forth is that there is no such thing as a baseline forecast unless you eliminate all other variables such as competitor promotions or competing product promotions.

Understanding that the above is certainly likely to be the case, we believe there is still value in baseline forecasts that use extended periods of time exceeding 12 weeks of non-promoted data in order to arrive at their conclusions. The reason for this is simple: By using a longer time series to determine the baseline forecast specific instances such as category cannibalization or competitor promotions will be reflected in the baseline forecast. The understanding that there are influences affecting even the baseline forecast becomes embedded in the forecast itself. For this reason, a baseline forecast that utilizes an extended data series will smooth out or negate the ripples caused by other competitive influences and still provide some valuable insight on how the product will perform moving forward in a non-promoted state.  Within ClarityRetail however, a baseline forecast is just a starting point as we will explain further below.

How Does the Traditional Demand Forecast Work?

Generally all forms of demand forecasting will look at historical sales data, remove the highest and lowest numbers then average it out over a reasonable period time. This means, depending on the period of time, promoted movement is mixed in with non-promoted movement. At ClarityRetail we believe that this approach is inherently flawed. Therefore, the mixing of the two types of movement in the demand data will inherently bleed additional error into the forecast being derived. In our experience working with grocery retailers we have determined based on empirical evidence that the difference between base demand and promotional demand can be so great that the correlation between the two is at best weak.  That is to say, in some instances a promotion can have such a dramatic effect on sales volume as to make the base demand forecast almost useless. In other cases, a promotion may have negligible impact on sales volume in which case the base forecast has increased utility. So the question then becomes how do we determine what the relative impact of the promotion is going to be?

At ClarityRetail we collect promotional data from your sales information and price feeds and manage it uniquely thereby enabling us to distinguish it clearly from the baseline forecast. We do still generate baseline forecasts as these are necessary when an item is not being promoted. ClarityRetail deals with this through a sophisticated set of algorithms that determines the best forecasting approach to use for a particular item’s behavioural pattern.

Demand-Driven Retailing – What Does it Mean?

A lot has been written about how retailers can become more demand-driven.  In this section, we want to talk about what it means to become more demand-driven and the tools you need to get there.  Without the basic building blocks you need to get started, you will ultimately hit a brick wall in terms of what can be accomplished.  Foundation therefore, is a key principle we will be discussing here.  Having the proper foundation in place positions you for success but only if you leverage what you have available and identify the gaps that need to be filled in terms of data and technology. 

At ClarityRetail, we believe that becoming truly Demand-Driven starts with accurate base-line and promotional forecasts using your daily POS data.  Your POS data represents the start of demand, the consumer.  By getting an accurate representation of consumer demand at store level, you can optimize your entire supply chain removing inefficiencies throughout.

The Importance of Accuracy - The Old Adage, "Garbage In, Garbage Out"

We cannot stress this point enough because we often find retailers who are struggling with getting accurate POS movement from their systems.  As stated, OOS-rates can have a significant negative impact on the accuracy of your POS movement data.  Sales of that item will not reflect true demand therefore leading to erroneous conclusions by any demand planning software including ClarityRetail.  Under the assumption that OOS occurrences are infrequent and irregular, ClarityRetail employs sophisticated filtering logic to assist in detecting OOS “noise” and other such similar anomalies.  In these instances, the sales data associated with the “noise” will not be used in forecast calculations.

In addition to OOS occurrences in POS movement, a retailer can also have product coding errors at the cash register or on package labelling.  The simplest and best way to deal with this real world problem in data quality is to increase the amount of data being analyzed.  The truth is that as more samples of good data are being analyzed, this will minimize the impact of the infrequent and irregular occurrences of bad data. Therefore, the rule of thumb is to extend the amount of data the forecast is being calculated from and this will help improve the quality of the forecast. Simply put, analyzing 4 weeks of sales data is not as good as analyzing 10 weeks for improved forecast quality.

Improving Retail Inventory Management Practices

One of the biggest challenges faced by retailers is managing inventory effectively at each store so that accurate on-hand stock levels are available for order planning.  For retailers with perishable commodities, this problem is further exacerbated.

ClarityRetail can provide you with the insights you need in terms of consumer demand so that you can reduce excess inventory in your stores and warehouses and maintain only what you need.

Inefficiencies related to inventory result in one of the largest expenses retailers incur as they grapple with the ongoing costs associated with excess inventory.  An accurate demand forecast helps you plan and execute better, enabling you to ship just the right amount of product to your stores.   Generally inefficient ordering practices lead to over-stocks as store level managers tend to over-order as a result.  On the whole, they would rather have more inventory than needed than have to deal with customer complaints, so over-ordering tends to be a common practice.   By empowering them with a forecasting tool that looks at a variety of influences on sales rather than a simple moving average over the last number of weeks, store operators can adjust order sizes appropriately.

Those Nasty Out-Of-Stocks

How many customers have visited your store and not found what they were looking for and ended up leaving dissatisfied.  In all likelihood, you will never be able to accurately come up with an answer.   Even if you have an effective out-of-stock tracking system in place, it can only give you an approximation of consumer dissatisfaction. 

According to studies of Corsten/Gruen (2002, 2008)^[1] the OOS-rate is about 8%. For products under sales promotion OOS rates up to 30% exist. Reliable information about demand is necessary for DCM, therefore lowering OOS is a main factor for successful DCM.

Corsten and Gruen describe key factors for lowering OOS-rates:

• data accuracy
• forecast and order accuracy
• order quantity
• replenishment
• capacity (time supply)
• capacity (packout) and planogram compliance
• shelf replenishment

ClarityRetail helps you reduce your OOS-rates in a number of the key factors cited here. 

Doing a Better Job on Promoted Items

We have written here about the value of a good promotional forecast.  With OOS-rates on promoted items hovering in the 30% range according to the Corsten/Gruen studies, it is imperative that retailers do a better job at planning promotions, starting with a good understanding of the promoted demand that can be expected to occur.  This is not as simple as looking at previous promotions of the same item.  Other factors that need to be considered include seasonality, current sales trends within the category, other items being promoted in the same category at the same time and of course, the price the item is being promoted at and the type of promotion.

Replenishment

Firstly, by giving you visibility to upcoming demand, ClarityRetail can be used to determine whether you currently have enough inventory in the store to meet it.

Secondly, ClarityRetail gives you a view of upcoming demand that you can use to optimize store ordering for the next shipment cycle.  Whether you’re on a centralized ordering system or allow store level personnel to create their own orders, ClarityRetail provides the visibility you need to order more effectively.

Getting the Data to the Right People

ClarityRetail reports can be sent to anyone in your organization who needs them.  Store-based forecasts can be sent to each store manager for their store each day helping them plan and order better.  If you are on a centralized ordering system, store by store forecasts can be sent directly to your central order management system.  By comparing the forecast to your on-hand inventory, you can quickly determine your days of supply and react accordingly.

Going Local – the Move Towards Localization

According to recent studies done by AMR Research, retailers in both Europe and North America have plans to add more localized content in their offerings over the next three years. In order to do this effectively, we must be able to identify store-specific sales patterns at the item level to determine the optimium assortment per store. Are you ready for the Going Local initiative you may be embarking upon soon? This growing trend towards local assortment as well as local procurement is creating new demands of today’s forecasting solutions. To work in a localized environment, they need to be adaptive to rapidly changing conditions. Depending on which retail segment they operate in, retailers must become more responsive to localized preferences. Demographic conditions based around ethnic make-up, income level, education level and other factors will determine not only the optimum product assortment for each store, but also what items and categories need to be carried and those you can live without.

Localization by definition demands that you start forecasting demand by store, category and item.