Order Right is an AI-powered, hyperlocal forecasting and auto-replenishment solution for diverse retail formats, including Food and Grocery; Health, Beauty and Wellness, and Convenience Stores.

Demand forecasting happens at the most granular SKU-location level. A multivariate forecasting approach is used to capture demand drivers, while a hierarchical modeling approach is used to address data issues at the SKU-location level.

Then, the best model is selected from an ensemble of candidate models, which consists of parametric models as well as ML and deep learning algorithms that forecast demand with hyperlocal accuracy for all kinds of SKUs.

The orders can be executed in three ways:

• Offline Method

  Recommended orders can be downloaded as an Excel file.

• Order Right to ERP

  Order Right’s output can be integrated with the ERP system for Purchase Order generation and order placement.

• Order Right to ERP to VENDOR LINK

  This is an extension of the above method where Algonomy’s Vendor Link (an AI-driven, end-to-end supplier collaboration platform) is integrated with the ERP for seamless P2P management with vendors.

Order Right supports the following order types:

• Direct Store Orders to suppliers
• Warehouse/Distribution Centre Orders to suppliers
• Stock Transfer Orders

Order Right has algorithmic methods, manual methods, and hybrid methods to handle new products.

Manual methods range from manually entering forecasts and replenishment plans to assigning similar products and corresponding percentages.

Algorithmic methods range from algorithmically determining similar products and algorithmically determining forecasts and replenishments based on the history of those products. This makes Order Right extremely effective and robust solution for retail sectors with volatile inventories, like health, beauty and wellness.

Order Right’s replenishment algorithm elegantly handles product discontinuations. It recommends orders so that inventory is gradually brought to zero on the planned phase-out date.

Of course it does. Its hierarchical forecasting framework starts with a collective forecast for similar products. This provides a rigorous mathematical framework to determine demand shifts as well as demand lifts due to reasons such as promotions and on-shelf availability.

Order Right uses a wide range of operational and contextual data to drive accurate forecasting and replenishment decisions.

Key data inputs include:

• Sales data, such as pricing, promotions, marketing, and related data, is used as predictors for demand forecasting.
• Inventory data, such as stock balances and expiration dates.
• Open orders that are not part of the current stock balance, along with ETAs.
• Minimum order Qty, UOM, Contractual lead time, Replenishment frequency, Order placement days (such as Day of the week/date of the month) to act as constraints in the replenishment algorithm.

No, Order Right doesn't need any PI data.

Ordering frequency is not recommended by the system. The user defines the ordering frequency, and the system recommends orders based on this frequency.

Order Right by default supports daily, weekly, or monthly forecasting frequency. It can be configured to support daily forecasts as well.

By default, the forecasting horizon is 12 weeks, but it can be easily extended through configuration.

It supports any product category that runs on a replenishment model (long product life cycle). This includes categories such as CPG products, Fresh, Bakery, Frozen, Beauty & Wellness, Specialty, etc.

Since Order Right is built specifically for retail demand forecasting and replenishment-related use cases, it has many inbuilt algorithmic techniques to handle noisy data, sparse data, outliers, and exceptions, which are inherent in retail data. It also has cutting-edge data imputation techniques to mitigate data issues.

Of course you can. There can always be exemptions where algorithms are not trained to handle. We believe in technology augmenting people and people augmenting technology!

Discounts and promotions are considered within the demand forecasting framework. We use multivariate forecasting that incorporates the impact of such factors to forecast demand. Since replenishment planning runs on top of the demand forecasts, it automatically captures the impact of discounts and promotions on replenishment planning.

Historical sales are recorded in terms of the selling UOM. Therefore, the Forecast is based on the selling UOM. In Order Right, there is a provision to configure the buying UOM as a multiple of the selling UOM.

As an example, if the selling UOM is a single unit and the buying UOM is a dozen, the corresponding configuration will be 12.

Accuracy is an outcome. The process of achieving high accuracy depends on two main factors: data and the algorithm. Order Right has an extremely flexible data layer where users can add any demand influencers without any pre-configurations. It automatically does feature pre-processing and feature engineering. This leads to a rich data layer, which is the basic ingredient of accurate forecasting.

On the algorithm front, it is important to use a wide array of different candidate algorithms. There is no “silver bullet” that works for all situations. Order Right is equipped with an ensemble of 2,000+ candidate algorithms, which ensures a good fit for all kinds of data.

While exact accuracy varies, it generally outperforms any traditional method by at least 20 percentage points.

Order Right is deployed in Microsoft Azure. Hence, data resides and processing happens on the Microsoft Azure cloud. We generally provision a single tenant for each customer.

Order Right does not need any technical expertise. While some technical expertise is needed to set up and configure the system, it is done by the Algonomy engineering team during the setup and implementation phase.

Algonomy provides extensive support on data-related issues, application-related issues, as well as output-related issues by data engineers, application engineers, and data scientists, respectively.