Frequently asked questions (FAQs) about Sterling Intelligent Promising

Find answers to frequently asked general questions that are related to Sterling Intelligent Promising.

General information

Is the clientId or clientSecret different for each tenant?: Yes.

How can I recover client credentials?: Open an account recover case with IBM® Support. For information about opening a case, see Getting help.

Does Sterling Intelligent Promising have a user interface?: Yes. For more information about the user interface, see Managing.

When I generate a unique token, what is the base64Encoded string that needs to be specified?

Use a base64encoder that is available and specify the input as : and encode it. A Base64 format token is created, which can be used for generating the bearer token. A Linux® command example is as follows:

>echo -n 'ABCD:1234'|base64
Results: QUJDRDoxMjM0

Can the invalid entries for both supply and demand be deleted?: No. Instead, you can start the Sync Supply or Sync Demand REST API by using the PUT request method, and set the value of quantity attribute to 0 (zero).

Can the reservation be updated by using REST API?: You can increase the reservation quantity for an existing reservation. For all other actions, delete the existing reservations and re-create them.

How is the 400 response code resolved?: The 400 response code indicates that the input is not correct. Verify that the mandatory parameters are provided in the input JSONs.

What response codes can retry on failure?

Any status code of 500 or greater can be retried.

If you receive a status code that is less than 500, see the REST API documentation for the corresponding API to understand what went wrong in the request.
Even though a request times out, verify that the request was completed successfully.
If it is an idempotent API call, for example, a PUT method, you can try again.

What is the rate limit for APIs for each environment?

A rate limit is the maximum number of API requests per hour. All lower environments, which include the developer toolkit, development and QA environments have a rate cap of 10000 API calls per hour. You also have access to a preproduction and production environment for which the maximum limit is defined according to your Service description contract.

To view your Service description contract, complete the following steps:

Click Welcome to IBM terms page.
In the Search field, enter the product name, IBM Sterling® Order Management to view your Service description document.
For more information about the API rate limit, see the Sterling Intelligent Promising Sterling Intelligent Promising Environment Limits on API Calls and API Actions section of the document.

You can upgrade the rate limit on the developer toolkit, development and QA environments. Alternatively, you can purchase a separate preproduction environment. To explore these options, you can book a meeting with IBM sales. For more information, see IBM Sterling Intelligent Promising.

Are there character limits in the Rules APIs?

Yes. For example, the Description field must not exceed 1024 characters. Exceeding this limit triggers validation errors. Reduce the amount of characters and update the rule.

Does Sterling Intelligent Promising support items with variations?: Yes. For more information, see Items with variations.

Does Sterling Intelligent Promising support unlimited or unplanned inventory?: No. As a workaround, you can provide inventory for that item at a node with a large quantity, such as 10000. You also need to sync this number periodically.

Does Sterling Intelligent Promising support vendor-managed inventory or Global Trade Identification Numbers (GTIN)?: No. Inventory is tracked at a specific SKU only and that SKU's availability only is inquired. Availability across SKUs is not supported. As a workaround, you can inquire for each SKU that shares the GTIN and aggregate the number outside of Sterling Intelligent Promising.

Self Service

Find answers to frequently asked questions that are related to Self Service.

Can an organization have multiple organization administrators?: Yes
Do I receive an email from IBM when new features are added to Self Service?: Yes, all users with the organization administrator, developer or stakeholder role receive product notification emails about the new features.
Does Self Service have the feature to control access to the production environment?: Yes. There are two different developer roles in Self Service, one for the nonproduction environment and one for the production environment to control access to the production environment.
Why don't I receive emails from Self Service?: Your mail server or spam filter might be blocking emails that are sent from Self Service. You can ensure that emails from oms.supply-chain.ibm.com are allowed in your mail server.

Optimization service

Find answers to frequently asked questions that are related to the Optimization service.

Is there an option to integrate labor management with SIP?: You can't integrate directly with labor management software. However, you can upload labor costs that are also known as processing costs for optimization.

Does the Optimization service consider physical weight or dimensional weight for each item in an order?: Thresholds are considered and then either dimensional weight or physical weight, whichever is greater, is used to determine the appropriate weight for consideration during the optimization decision process. Carrier-applied thresholds for dimensional weights are managed within the platform.

How long are orders retained for optimization explanation?: Orders are retained for optimization explanation for 90 days.

Can I activate the optimization levers gradually?: Yes, you can enable the optimization objectives incrementally. You must activate the shipping cost optimization lever from day one. Then, you can activate more levers over time as the relevant data is configured in Sterling Intelligent Promising.

Sterling Intelligent Promising Premium

Find answers to frequently asked questions that are related to Sterling Intelligent Promising Premium optimization with AI.

AI and Machine learning (ML)

Is the ingestion of historical or future data for markdown and stockouts optimizations to accelerate machine learning supported?: Yes, the ingestion of historical data to ramp up the machine learning phase effectively is supported.

How soon does the model start to deliver business value? Does it need to run for a few months first: The model requires three months of historical data at least to begin training. When it is trained, it delivers measurable value.

How is the model improved continuously over time?: The model is retrained monthly by using data that flows consistently. This allows the model to adapt to changing patterns and to improve accuracy over time.

For advanced models, where files are pushed to IBM Cloud® Object Storage, is there any visibility into how model predictions are calculated?: Yes, while the demand forecast is not shown, you have visibility into the cost metrics that are derived from the predictions. The Optimization service converts the forecast data into a cost. The costs are used to decide about optimizations.

Can the data feeds vary by enterprise or location, such as the US and Canada? Accordingly, does Sterling Intelligent Promising tailor the optimization benefits?: Enterprise-specific data feeds and the filtering of benefits is not currently supported.

Can I activate the business levers separately? For example, can I start by enabling markdown optimization before I enable stockout optimization?: Yes, you can activate the business levers one at a time, allowing you to start with a specific use case and then to expand gradually.

If T-log data only is available and markdown data isn't available, can I activate the stockout optimization lever?: Yes, you can activate the stockout avoidance lever without markdown data.

Can I activate the optimization objectives gradually?: You can turn on the business objectives one at a time. You must enable the shipping cost optimization from day one. Then, you can add the other optimization objectives as and when the data is configured in Sterling Intelligent Promising.

How much manual intervention is needed to run the AI models daily?: When the data integration is set up, no manual intervention is required to run the AI models daily.

Can the model training be triggered manually?: Yes, but you need to contact IBM support to do this. For more information, see IBM Support.

Which type of data is required to train the AI model?

T-log
Markdown
Order line status
Catalog item
Availability
Supply plan

Which data integrations are built in and which require custom development?

Inventory availability and supply data are integrated automatically between the inventory service and the optimization service.
Catalog data is integrated between the catalog service and the optimization service.
For IBM, the order line status data is integrated with the optimization service.

What kind of forecasting models are used to calculate the components to avoid stockout and markdown risks?

Feature set
Forecasting duration
Specific ML model

Seasonality is used to train the models. Data from the past year is used to identify seasonal trends and current three-month data is used to train the forecasting models. The models employ regression and deep learning techniques to forecast daily in-store and Buy online and pick up in a store (BOPIS) sales. Key model features include statistics, computed price, historical inventory and historical velocity.

What is the time horizon that is used to calculate stockout and markdown risks?: A 60-day horizon is used to calculate the probability of stockouts or markdowns.

Can model predictions be improved by providing datasets for a longer duration?: Yes, extending the dataset duration improves both the accuracy and the longevity of the models. For optimal model performance, ensure that data integrity is maintained between datasets when you feed in datasets for extended durations.

How is the accuracy of the model tested?: The model is tested for model bias and drift, which might be introduced as a result of missing data or changes to the data values.

How frequently is the model monitored to help ensure that it stays on track and how often is it retrained?: The model is trained once or twice a month by using the most up-to-date data. Then, model serving occurs daily and often in real time to help ensure that forecasts are updated for every event change. At a minimum, forecasts are refreshed daily.

Is there a data feed to improve the accuracy of the model?: Shipment data enhances the ability to plan for a better feedback loop. The models produce exceptional results currently. However, you can incorporate the shipment data into the new node reliability models to identify differentials to improve the model further.

Does the Optimization service generate the relative velocity of items?: Yes, forecasting is done for each product at each location. Then, the relative stockout or markdown cost calculations are based on the inventory levels.

What happens when a newly released item is still not sold?: If an item wasn't sold by a store in the past three months, a statistical value is borrowed from other similar items.

Where is the replenishment data used?: The replenishment data is used to determine the incoming supply rates that help to assess the forecast and stockout probability. Then, based on this information, the stockout cost is calculated.

Can I see how orders are optimized by using markdown and stockout avoidance models in Sterling Intelligent Promising: Yes, you can see how orders are optimized by using markdown and stockout avoidance models by signing up for the Sterling Intelligent Promising trial tenant. When you add predictive models, such as markdown and stockout avoidance to your optimization objectives, you can see how they impact order decisions and review the associated benefit report. For more information, see Register for your trial.

Markdown data

For markdown avoidance, how do I prioritize the slow-moving products when the rate of sale is zero or close to zero?: You can use artificial markdown to prioritize slow-moving items.

Are corporate, store and banner level types of markdown supported?: No, these banner types are not supported.

For the markdown algorithm, which one of the following values is most predominant:

Transaction total retail amount with discounts
Transaction total charged amount

The markdown item retail price is not mandatory? How does the model calculate the benefit without this price?: The model calculates the benefit by using the markdown percentage. In the markdown feed, either the markdown price or the markdown percentage is required.

The markdown event ID is mandatory. How is this ID used in the model?: The markdown event ID is used to uniquely identify and track individual markdown events during markdown predictions.

Why is an end date not required for a regular markdown, but required for an artificial markdown?

For regular markdowns: A markdown item remains active until the status is changed to MKDOWN-DELETE, eliminating the need for an end date.
For artificial markdowns: Since artificial markdowns are temporary, an end date is necessary to define their duration and help ensure proper tracking.*

If both markdown retail price and markdown percentage are provided in the data, how does Sterling Intelligent Promising calculate the net retail price after markdown?: Sterling Intelligent Promising uses the markdown retail price to calculate the net retail price after markdown. The markdown percentage is not considered if markdown retain price is provided.

T-log data

When calculating the probabilities of stockouts, which value takes precedence?

Transaction total retail amount with discounts
Transaction total charged amount

The transaction total charged amount is given higher precedence in the stockout calculation.

When dealing with exchange transactions, can the data be sent as two separate transactions, one for the return and another for the sale?: Yes, exchange transactions can be sent as two distinct transactions, with one representing the return and the other representing the sale.

In a T-log file, if an item is stamped with a discount and a markdown transaction item status code, which of these codes takes precedence?: An item can have either a discount or a markdown status code only stamped but not both. As a result, there isn't any precedence.

Supply plan data

Where is the supply data used?: The supply data is used to determine the incoming supply rates to assess the forecast and probability of a stockout. Then, based on this information, the stockout cost is calculated.