ForecastPlanner™ enhances and optimizes the forecasts within both SAP APO and SAP ERP forecasting components. It functions as central forecasting hub that can service many of the SAP modules, especially APO DP. Master data and time based data is received directly from the SAP infrastructure and the aggregation/disaggregation according
to time or other characteristic can be automatically executed. Thus, a hierarchical forecast is possible even in places that do not support it naturally (e.g. MRP).
and other systems
ForecastPlanner™ input can be a table view, an Excel™ or csv file containing the data in its raw format. An XML template controlling the data interpretation allows a hands-off operation. The execution of the session includes forecasts and smart optimizations that detect the best forecast models and algorithms for the targeted business scenario. The results are sent to the hosting system and are also available via the web site analysis screen. The configurations and results can be exported to the user's desktop as an Excel analysis cube and report.
Multinational corporations need to apply a centralized Demand Planning process into several locations. That requires a uniform solution across the different markets and business units although they may widely differ in various aspects. However, adopting the Demand Planning model to each of them is an expensive task.
ForecastPlanner™ provides an easy, cost-efficient way to provide re-tuning and maintenance of the Demand Planning process. It enables the export of market characteristics, sales & forecasting while identifies stable Demand Planning methods that will adopt to the market.
Events and Promotions
ForecastPlanner™ allows users to connect seemingly "one off" events into a cluster of events with predictive behavior and to model "floating seasonality". These events are automatically filtered out, forecasted separately and recombined with the baseline data at the appropriate places. ForecastPlanner™ includes any supporting information to understand the reaction to price points, discounts and events behavior while using past demand, past revenues and advertisement activities.
• Intelligent acceptance/rejection of causal information.
•Computes price elasticity functions for different scenarios.
• Analyzes the economic value of a discount to find the optimal price point.
HOW IT WORKS
ForecastPlanner™ is a highly advanced automatic forecasting, inventory & business analysis optimizer and full-cycle demand planning solution, providing the uppermost attainable accuracy.
The cloud-based SaaS solution utilizes any available information, including multiple data sources, hierarchies, product life cycle, noisy and promotional data sets to create ‘best of breed’ forecasts.
Ogentechs turnkey solution provides immediate results, quick return on investment and low cost of ownership.
ForecastPlanner™ is easily integrated and requires only minimal manual intervention. It operates as a standalone solution or can be deployed to optimize the settings of Demand Planning systems like SAP, APO.
No need for data replication to another system, which results in a reduced interface complexity and maintenance need.
The hybrid engine continuously optimizes models, hierarchies, policies, algorithms and parameters according to the user’s specified business objectives. The solution is fully scalable and is customized precisely to the user's needs.
For attaining the most accurate forecast results, our smart algorithms find the best data models and combine all relevant hierarchies, price elasticity, promotions, events, life cycle management, cannibalization, sentiments, and revenues.
Each target horizon, resolution and combination level can be forecasted separately using differing cleansing methods, algorithms or parameters.
The frequent exceptions detection, error calculation methodology and re-optimization of the paradigm, ensures maintaining accuracy optimization and sustained bottom line results.
Ogentech developed an innovative noise reduction methodology for different scenarios, revealing the exact item and price elasticity to create profitable promotion forecasting even when the past information is sparse.
Events are automatically filtered out, forecasted separately and recombined with the baseline data at the appropriate places.
Demand data can be specified in multiple granularity levels as well as multiple time resolutions.
ForecastPlanner™ interacts with the planner's knowledge to create a detailed business overview, which is presented in customized and user-friendly Power BI reports.
Those valuable business insights have a great impact on marketing, management and supply chain related operations as they provide a complete business overview.
The exceptional forecasting quality and process automation are based on the innovative optimization infrastructure such as the smart optimization solution which is designed to find and combine the best data model, time base, parameters and algorithmic combination that will lead to the optimal, consistent and best results over time. According to specified business objectives, ForecastPlanner™ computes the right time and hierarchy based aggregation and disaggregation, methods of outlier correction and filtering as well as algorithmic methods.
ForecastPlanner™ utilizes any available information, including multiple demand data sources (orders, shipments, POS, invoices), catalogue and sales hierarchies, product life cycles, promotional information, revenues, client preferences, and partial information data sets – all to create the most accurate and robust forecast, promotional elasticity and target inventory planning. At every run, ForecastPlanner™ imports the entire set of data and a fresh catalog:
A phase of learning and detailed data analysis precedes the planning and allows the system to assess the data properties and noise level for each forecasted node.
Data is sliced and segmented to create a very fast yet intelligent execution cycle.
ForecastPlanner™ flexible engines optimize models, hierarchy use, policies, algorithms and parameters at every run according to the user’s needs and recent catalog & data changes.
For attaining the most accurate forecast results, the system always searches for the best match of data, cleansing (statistical and promotional) and hierarchical forecasting.
The algorithms are individually adapted to each forecasted node thus allowing seasonality and trend to be derived from the catalog hierarchy, promotional uplift from the sub chain, noise cleansing from the daily material @ the store information and forecasting from the weekly material @ store sales data.
The valuable analysis of business insight allows a highly advanced promotion and holiday planning that will greatly increase your revenue. Our solution attempts to deliver highly improved result while breaking the complexity of the forecasting process. Thus, allowing the users to focus on the truly important issues which require human attention.
The target users include budget and revenue planners, HR planners, CRM, SCM/SCE, traders, demand planners, and financial forecasters.
Noise in data consists of any unexplainable phenomena and often interlaces with trend, price elasticity and seasonality. Hence, noise can be interpreted as the start of a trend or as seasonality.
Heavy-handed noise cleansing can reduce/eliminate the effects of seasonal and trend and result in trivial or wrong forecasts. In order to prevent this, ForecastPlanner™ has a special noise analysis and detection module using segmentations, local and global windows, hierarchical information and pattern recognition to detect and filter noise. Ogentech developed an innovative noise reduction methodology for different scenarios, revealing the exact item and price elasticity to create profitable promotion forecasting even when the past information is sparse. Events are automatically filtered out, forecasted separately and recombined with the baseline data at the appropriate places. Demand data can be specified in multiple granularity levels as well as multiple time resolutions.
The basis for the automatic calculation of future lifts is based on Promotional Clustering. Furthermore, it sets the basis for analysis and discussion prior to fixing a price or deciding on a promotion. ForecastPlanner™ calculates the expected sales for a future price discount and proposes the lift level based on provided future price and supporting information.
Samples that do not fit the worldview should be rejected, as newer samples are preferred. However, the collection of rejected samples may create an alternative world view.
ForecastPlanner™ Clustering generates higher sales by:
• Stronger activities,
• Better presentation/store salesperson bonuses,
• Discounted prices.
A key component of ForecastPlanner™ is the Stabilizer. Many Demand Planning systems are incapable of controlling the accuracy of the forecasting quality over time because systems, that are merely predicting a flat demand will inevitably fail during changes of seasons. The aim of our stabilizer is to identify and deploy only the Demand Planning models that will produce good results over time.
ForecastPlanner™ is capable of running highly complicated structures with millions of combinations. The scale-up complexity is almost linear in the number of nodes and does not explode with the growth of the client. A full optimization and forecast run of a client with around 10K nodes runs at about 4 minutes on a standard office workstation.
Retail Value Proposition
• Improvement in accuracy.
• Noise detection & cleansing for the creation of conditioned past and better future baseline.
• Real elasticity tables and promotion planning.
• Fast learning during a new product launch.
• Accurate computation of target stock for multiple scenarios.
• Improvement in availability.
• Reduced rate of stock outs, dead stock and spoilage.
• Reduction in manual labour.
• Better flexibility and quicker reaction to changing behavior.
• Fast exception detection.
Creation of Target Inventory
• Target inventory is based on the item forecastability and on intelligent detection of past behavior.
• Target inventory is calculated differently for every item/store based on its segment (steady, sporadic, short, seasonal, etc) and on its life status (launch, promoted, phasing out).
• Management of promotion target (how to start and leave a promotion).
• Forecastability and variability of demand and supply govern the safety stock portion of the target inventory.
• A careful past behavior pattern recognition verifies/updates the target inventory.
HUMAN & STATISTIC INTERACTION
To attain a maximal forecast accuracy, we combine statistical knowledge with the planners ‘know-how’.
This way, the forecast quality profits from both, the precise statistic proficiencies of our algorithms, complemented by human qualifications such as experience and supplementary knowledge. By pointing out items which require human intervention, ForecastPlanner™ enables the user to be more responsive to market needs and to concentrate on data enrichment, occurring challenges, cannibalization, and other non-statistical properties.
Managers, planners, and analysts can request customized Business IntelligenceI reports as well as hands-on ‘field’ data for various business levels which enable a full data overview and introduces a dynamic dimension of control over each operation unit.
Our user-friendly, one screen Power BI interface report offers a simplified exception detection, which can be adjusted to any set of conditions at any time. On grounds of our long pedigree in the field of demand planning, we developed an Advanced Alert tool that indicates and separates items which require human attention. This differentiation allows planners and analyst to focus their expertise on relevant issues.
ForecastPlanner™ is capable of clustering decisions and policies according to their eventual success in reality. By being able to combine results from different time resolutions and histories, it identifies not only optimal but also robust data models and configurations that provide consistent and best results over time.
ForecastPlanner™ enhances and optimizes the forecasts within other forecasting systems. Our solution can either be operated in a batch mode under existing Demand Planning systems & forecasting modules or function as a stand-alone solution.