The company has the ability to measure and report on the impact of promotions on sales, margins, and inventory to evaluate their effectiveness. Ideally, more complex metrics are available that capture:. Industry Challenge : While most retailers have some mechanism for reporting on promotional effectiveness, these typically require a great deal of human intervention to pull numbers from disparate systems and manually create reports.
Very few can take advantage of the kinds of advanced statistical analysis that can produce the more complex and meaningful metrics. Merchandise plans incorporate marketing strategy and promotional space allocations and accurately reflect promotional activity. Industry Challenge : Promotional plans are frequently created on a shorter timeframe than merchandise plans.
This means that merchandise plans are complete and orders placed prior to promotions being finalized. Past performance of promoted products or model-after items for new products is used to predict the lift on new promotions and the impact on the overall category.
Forecasting models take into account sales history as well as causal factors such as pricing differences, channel differences, duration changes, etc.
Industry Challenge : Very few retailers are adept at accounting for promotions in their inventory management activities. Most retailers have solid business processes and system solutions for some of these capabilities, but virtually none can boast of being strong in all of them.
If a retailer struggles with any of these elements, then corrective investment is warranted. Looking beyond this foundation, there are exciting developments that allow retailers not just to predict the results of promotions, but to model and optimize promotions in order to attain desired outcomes. This advanced functionality is in its infancy, and while at this point few software providers can demonstrate currently available offerings, the handful that do exist are extremely promising.
These applications combine data from multiple sources, including marketing, CRM, e-commerce, merchandising, planning, and finance. Collecting this data and making it available for optimization is no doubt a significant effort, but the resulting performance improvements can be astounding.
Consider the amount of time and effort retailers put into planning, managing, executing, and measuring promotions. Add to that the amount of margin wrapped up in promotional markdowns, and ultimately, even slight improvements in the effectiveness of promotions can drive substantial business results. Beyond quality product, enticing assortments and engaging shopping experiences, when considering pricing and promotions, these expectations include clarity of messaging, competitive pricing, product feature information and available inventory including specific locations.
What can retailers do to confront these challenges and effectively manage promotions and related demand signals? We offer the following recommendations:. While there are sophisticated tools to assist retailers today, nothing can replace the eye and heart of a strong merchant. Fundamentally, retailing is still all about the product, and leading processes and policies should be driven by the initial brand strategy and reflect the marketing strategy.
Consider the science. This is where retail leaders are truly starting to differentiate themselves. Leverage the best available market data, along with historical performance, aligned with the master sales and event calendar, to generate a predicted sales demand, as well as options and scenarios for promotions and promotional impact. Managing the science can be an art itself, a tricky orchestration of configuration, input gathering, data preparation, analysis, and interpretation—supported by advanced analytical solutions and appropriate skillsets.
Solidify basic blocking and tackling. This balances the art and the science with execution. Retailers must deeply consider and design key execution tactics to successfully determine, communicate, deploy, and process promotional transactions. Included here are back-office merchandising management and marketing, as well as in-store execution.
These, coupled with channel convergence collectively address highly varied consumer shopping behaviors. Essentially, retailers must identify and communicate the right promotional messages to the best target consumers—at the appropriate time and in channels that will resonate—and then complete the circle with efficient transactional execution and informed, consistent customer-facing touchpoints. This gives store staff sufficient lead time to build their in-store displays and stock shelves, ensuring availability during the high-demand promotion period.
The system then uses information on local sales, demand forecasts, available stock in distribution centers, distribution center capacity, and other factors to fulfill the remaining demand following standard replenishment cycles.
To improve automation, retailers can develop templates for different types of stores and promotions to fit their business characteristics and goals. These templates are used to quickly implement configurations that automatically calculate initial store orders and subsequent replenishment orders—even ramping down toward the end of the promotion period if desired.
Retailers who bring forecasting and replenishment planning for stores, online channels, and distribution centers into a centralized system ensure full supply chain visibility.
For example, elements like forecasted customer demand per store, fulfillment channel, day, and product—when combined with replenishment schedules and batch sizes—can be used to more accurately calculate end-to-end inventory needs. Whether running promotions in stores or through multiple sales channels, retailers can leverage this supply chain visibility into improved promotion management.
In an integrated supply chain, DC forecasts are based on store order projections, which ensures that the demand impact of any planned promotions is made immediately visible throughout the supply chain Figure 5. This helps planners secure the right amount of inventory for the right locations on the right dates. Omnichannel retailers who run promotions both in stores and online face an additional layer of complexity in promotional forecasting.
However, they can streamline that complexity with an end-to-end system that also maintains separate forecasts by channel. This kind of visibility and integration can also be used to ensure online availability through virtual ringfencing. Virtual ringfencing uses online demand forecasts to automatically reserve inventory in your distribution centers, preventing store pre-orders from claiming too much stock.
With access to channel-level data on promotional sales, the system can also automatically adjust or remove ringfencing rules as needed. End-to-end inventory management also benefits the operative buying teams responsible for efficiently exploiting rebates to improve gross margins.
In theory, smart buying when suppliers offer temporary price reductions for promoted items is quite straightforward: retailers should order less product just before a temporary price reduction and stock up when the price is low.
Storage space availability and product shelf-life also need to be factored in. To optimize forward buys, a planning solution should consider these restrictions alongside time-dependent trade promotion data. Supply chain integration gives the solution access to all the information it needs to automate calculations for when and in what quantities orders should be placed.
Ultimately, end-to-end inventory management enables retailers to plan promotions once, then automate execution based on forecasted demand and inventory requirements throughout their supply chain. The results can be significant. Retailers who target improvements to promotional forecasting and replenishment typically also see benefits like optimized delivery flows and more efficient capacity utilization.
As retailers compete for consumer attention and customers embrace the convenience of omnichannel shopping, retail promotions are only growing in importance and complexity. More information is available in the Forecast Administration workbook.
This measure displays the promotional forecast where the promotional peaks are not calculated by RDF, but imported as lift overrides. Override, in units, of the system generated promotional lift. The override is at the same intersection as the forecast. This measure can be equal to the Aggregated Promotion Lift Override or zero 0. If the indicator is set to True which applies the override , then this value should be the same as the spread down Aggregated Promotion Lift Override.
If the indicator is False which does not apply the override , then this value equals zero 0. This measure displays the total effect that a promotion can have on an item's sales units. This measure displays the total effect that a regular price change can have on an item's sales units. The Promotion Parameters view allows you to set up rules that need to be followed when forecast is simulated.
For instance, you may not want to apply regular price change effects when an item is on promotion or you may want to incorporate Halo and no Cannibalization effects in the forecast. This measure allows you to override in units the system generated promotional lift.
The override is at an aggregate level and spread down to the forecast level. Select this parameter set to True if the application of Cannibalization Lifts is enabled for the forecast level. Select this parameter set to True if the application of Halo lifts is enabled for the forecast level.
Select this parameter set to True if you want RDF to ignore the system calculated lifts and apply the overrides. Select this parameter set to True if the application of Regular Price Cannibalization Lifts is enabled for the forecast level.
Select this parameter set to True if the application of regular price Halo lifts is enabled for the forecast level. Select this parameter set to True if the application of regular price lifts is enabled for time periods that are promoted. If the parameter is clear, then regular price lifts are not applied for periods with promotions.
Select this parameter set to True if the application of regular price lifts is enabled for the forecast level. This measure is registered for each promotion. This is the value for the promotion effect that was used to create the forecast if no override is specified. Depending on the administration and maintenance settings it can be one of the following:. A combination of the two.
The way the two effects are combined can also be determined by the user by specifying a blending parameter. Causal variable types define how causal variables are treated in the causal model-fitting process which includes a call to the lower-level regression engine and the forecast generation process where the model is used to extend the forecast over the forecast horizon.
This read-only measure allows you to override in units the system generated promotional lift. The System Calculated Effect is a read-only measure indicating the lift effect generated by the system. The user-specified lift effect. Otherwise, it equals 1. In order to correlate deviations from the seasonal forecast with the occurrence of historic promotion events, the system needs visibility as to when these events were active.
The system must also be informed of dates on which the status of upcoming promotion events will again be on, so the anticipated promotion effects can be built into the forecasting model. The Promo Planner task allows you to indicate to the system when certain events were active in the past and when they is active in the future. All promotional events should be represented as accurately as possible so the modeling routine can more precisely detect correlations between event occurrences and changes in sales values.
The Promo Planner task consists of as many views as are necessary to represent all unique dimensional intersections associated with the promotion events contained in the task. A separate view is constructed for each of the required intersections. In this setup, the Advertisement and Gift with Purchase promotions would appear on one view, and Christmas would appear on another. Whenever a hierarchy is not included in the base intersection as in the case of the Christmas promotional event the event is assumed to apply to all positions in the undefined hierarchy.
Thus, Christmas is assumed to apply to all products and all locations, but only to the Day-level calendar positions specified in the Promotions View. The Workbook wizard opens. Select the promotion events to include and click Next. This view provides an interface in which you can specify the time periods and possibly products or locations for which certain promotional variables are active. The Data Type includes Boolean and Real types. The Model Type includes Linear and Exponential types.
A typical example of a metric modeled as exponential is the percent price change. Note how the percent price change can be positive price increase or negative price decrease. Exponential promotions can be enabled together with Boolean promotions at the same time. However, they can not be enabled together with Real promotions. A value of 0 indicates that the event's status is off. A value greater than 0 will act as a weight when calculating the effects during the stepwise regression.
Real promotions can be enabled together with Boolean promotions at the same time. However, they can not be enabled together with Exponential promotions. Among the ways Causal variables can be implemented include price or discount percent.
Your Oracle Retail Consultant can best determine the most accurate setup of promotion variables based upon your promotional forecasting requirements. The Promotion Management task allows you to enable and disable promotions for causal forecast levels and to specify whether promotions can have a negative effect. The Promo Variable Model Type view allows you to view the model type for the selected promotion variables.
The Promotion Enable view allows you to enable or disable a promotion for the causal forecast levels. The Promotion Enable view is built at the hierarchy intersections of the promotion variables and the causal levels selected during the wizard process.
The Enable Promotions measure allows you to enable a subset of promotions for a certain causal forecast level. It defaults to True for all causal forecast levels.
Each promotion's type is defined in the Configuration Tool. For additional information, see "Promotion Variables". The Accept Negative Lift view allows you to specify whether a promotion is allowed to have negative effect. The Promotion Enable view is built at the hierarchy intersections of the promotion variables selected during the wizard process. The Promotion Allow Negative Lift measure allows you to specify whether a promotion is allowed to have negative effect.
It defaults to True for all promotions. The Promo Effect Maintenance task provides a view to the system-calculated and adjusted lift effects. The Promo Effect Maintenance task contains one view. There may be multiple versions of this view, defined at various causal levels. Select the causal final level and click Next. The Promotion Effect Parameters view allows you to view and modify the system-calculated effects of a given promotion at the causal source level.
There is one view for every causal source level. Table lists the options. The inclusion of the Promo Effect is decided by regression. If the Promo Effect is found to be significant on the training set, it is included in the model.
Otherwise, it is rejected. Automatic is the system default Promotion Effect Type. The Promo Effect is forced in to the model, thus regression is not given a choice to reject even if the effect is considered insignificant by regression. As a result, we will always return an effect even if it has a negative impact to the demand forecast.
This type allows you to specify a causal effect that is used during the forecasting process. The causal engine de-causalizes the training data using the user-specified effect. During forecast generation, the user-specified effect is used to determine the causal forecast. Therefore, you must change the Promotion Effect Type when this user-specified effect is no longer to be used.
The calculated effect is not written back to the effects measure, but it is used to de-causalize the data. During forecast generation the calculated effect is ignored, and, instead, the user-specified effect is used to produce a causal forecast. This promotion type is used in conjunction with the Causal Higher Intersection set in the Forecast Administration Workbook.
If the Causal Higher Intersection is not specified, no promotional effect is calculated. The way the system handles this, is by having the override effects measure filled in with higher-level effects for those variables specified as Override Higher Level. During forecast generation the effect calculated as average of low level effects is used to produce a causal forecast.
If the promotion variable is always be set to 0. The Promotion Parameters view allows you to view and modify the system-calculated effects of a given promotion. Select the specific products that you want to view and click Next. These measures, numbered , allow you to flag up to 20 float events as active in history or to be active in the future. In this task you can perform every step necessary to plan promotions.
You can determine which events should be active, the value of the price discount, review and possibly overwrite causal effects, and run simulations to evaluate the plan. The simulations are a very useful tool to determine how effective a promotion is. Once you decide on the setup for the promotion, based on the what-if simulations, the information can be committed and the next forecast batch generates the baseline and lifts that are exported to downstream applications, such as replenishment solutions.
In this step, you can review all the information related to promotion effects. For example, you can check the calculated effects at the final as well as pooling levels. You can decide if the calculated effects need to be adjusted, and how to combine final level and pooling level effects to create a robust causal forecast.
The views in this Promotional Parameters step give you the tools to build the blocks necessary to build a robust and accurate causal forecast. This parameter sets the weights for combining the Final and Source Level promotion effects, when calculating the blended effect. The range of the parameter is 0 to 1. A value closer to 1 will yield a blended effect closer to the Pooling Level effect.
You pick a high value if you want t robust causal forecast, although the final level promotion information is not very reliable. A value closer to zero yields an effect closer to the Final Level effect. You pick a low value if the final level promotion information is accurate and you want the causal forecast to reflect item-specific effects.
The measure can also be edited in the Forecast Administration workbook, and can be overridden in the Forecast Maintenance workbook.
The list of values displayed in this field allows you to change the pooling level that is used for calculating the blended effect. A value from the pick list is required in this field at the time of forecast generation. This Default Overlapping Promotion Adjustment Factor specifies at a high level how the individually calculated promotions interact with each other when they are overlapping in the forecast horizon. This parameter serves as a global setting, but can be overridden at lower levels.
The default value is 1. A value greater than 1 means the promotion effects will be compressed when applied in the model, instead of linearly summing up to get the total promotion effect. The larger the value is, the larger the compression effect will be, meaning the smaller the total effect will be.
A value between 0 and 1 means the promotion effects will be amplified when applied in the model, instead of linearly summing up to get the total promotion effect. The smaller the value is, the larger the amplification effect will be, meaning the larger the total effect will be. If a value less than or equal to 0 is put in the cell, the calculation engine will find the best adjustment factor to fit the history data.
This factor is then used to combine overlapping promotions in the forecast horizon. With few exceptions, the value should be anywhere from 1 to 5.
Additionally there are a few measures shown in Figure that when used together with the ones in Figure give total flexibility to you to generate the causal forecast you want. All of these measures are described and are available for editing in the Forecast Administration workbook.
0コメント