Predictive Ordering

Predictive Ordering Formula

For non-catch weight items 

  • (Predicted usage * days until order) + Floor - Quantity on-hand - Quantity in outstanding unreceived orders - Quantity in received orders = Quantity that needs to be ordered.

For Catch Weight Items

  • (Predicted usage * days until order) + Floor - Quantity on-hand - Quantity in outstanding unreceived orders - Quantity in received orders = Quantity that needs to be ordered.

  • The formula is the same as above but the final Case amount populated as the Quantity that needs to be ordered is " the quantity that needs to be ordered/ Bid File Catch Weight” Decision Logic will round up to the nearest full case. 

  • Predictive Ordering will calculate how many cases are needed for each catch weight item.

For Broken Cases

  • Items set to BOTH that have been ordered and/or received during the window used for the calculation will be an average of any EACH and CASE quantities

Predictive Ordering Guide Setup 

Enable Feature

Prior to rolling this feature out, you will want request that we enable it on your Decision Logic website by submitting a support ticket to the Decision Logic Help Center

Predictive ordering is set up on a  Distributor basis for both Manual and Auto types. 

To turn on predictive ordering for a particular distributor the process starts on the Distributors Maintenance page. 

Navigate to Maintenance > Distributors

  • On the Distributors tab first, select the distributor you wish to allow predictive ordering for by checking the Enable Order Prediction (Stores) checkbox on the right-hand portion of the screen.

  • By checking that box the Order Prediction Methods area will become active.

  • Select the type of methods that you want to be made available for prediction by checking the box next to them. 

Prediction Methods

Ideal Usage

  • Based off 8 weeks of historical Ideal usage for an ingredient and removes any values outside of 2 standard deviations before getting the average.

  • subtracts the last inventory and - x from orders that are not received and then subtracts any that were received.

  • This method requires an order scheduled to be set up for the distributor and will only properly predict usage for items that have a full recipe build out.

Actual Usage

  • Based on 8 weeks Actual usage for an ingredient removes any values outside of 2 standard deviations before getting the average.

  • This method requires an order scheduled to be set up for the distributor and will only properly predict usage for items that are inventoried on a weekly basis.

(warning)  This tracks the number of orders since an item has been ordered and will prompt the user after a set number of order to order a product.  This method can be used for any type of item.

Total Sales, Food Sales, and Bar Sales

These methods track the dollars of sales since an item was last ordered and the projected sales till the next available order and suggests orders based on a Case per X dollars of sales basis. 

Example 1 Case of To Go Boxes for every $40,000 in total sales)

Example 2 say on 8/1 you order Paper Towels, & you now have $0 in sales since the last time that was ordered. Say three days go by and you still haven't ordered it. You would now have something like $16,000 in total sales since it was last ordered

  1. if your floor set up on the order guide called for a case every $10,000 then the order prediction would tell you to order 2 cases (1.6 ~2)

  2. when you order it, the Total Sales count goes back to $0

  3. if you made that floor $300,000 then it would take weeks to get there before it showed up as a predicted item

Order Schedule 

To continue the set up click on the Order Schedules tab on the distributors maintenance page. 

On this page, you will set up the schedule for when your stores order from the distributor. 

  • Click the NEW button to create a new schedule. 

  • Name the schedule in the Schedule Name field on the right and then select the days that the stores place orders by checking the box next to each day.

  • After selecting the days, the Delivery Delay (Days) field is used to show how many days after the order is placed that the store will receive delivery of the product. 

(star)Note, you can set up multiple schedules for different ordering cycles for different stores. The stores will be assigned to the schedules in the next step.

After creating the schedules click on the Schedule Stores tab to assign stores to the various order schedules. 

  • Select a Distributor and the schedule you wish to assign from the drop-downs at the top of the page

  • Then select the stores you wish to assign to the schedule and move them into the Assigned Stores box on the right by using the arrow buttons between the two boxes.

Select Schedule Detail to generate the days of the schedule by month and date

You can generate the projected schedule for any month and then make modifications to the schedule.  To make a modification for a date, select the order date from the detail below and it will allow you to make your changes and add comments. In the case that you need to make one time adjustments to the schedule for occurrences such as holidays, the Schedule Detail tab was created. 

  • Define Distributor, Year, and Month. 

  • Select an Order Schedule from the Left

  • Select Generate to load the dates of that scheduled. 

  • To modify select an order or receive date, adjust and save.

(star) Note, the stores can also place OFF CYCLE orders from the normal Order Entry screen so they can have correct predictions even if the holiday schedule isn’t modified for them on this maintenance screen.

Assign Prediction Methods to Bid File Items 

The next step is the process is to set up the actual prediction methods and parameters for each product

This is process is started on the Bid file Maintenance page.

Navigate to Maintenance > Bid Files

  • Select the distributor that you are working with on the bid file page,

  • Select the Order Predictions tab. 

  • For each distribution item that you want to turn predictive ordering on for, select a prediction method from the drop-down on the right. 

(star) Note, not all items from a bid file have to be set up for prediction.  It is generally a good practice to start out with some items that are consistently ordered and let the stores get comfortable with the predictions and then slowly increase the items that are predicted.

Assign Prediction Values and Floors to each Order Guides

After setting up the prediction methods go to the Order Guide maintenance page to set up the parameters for the predictions.

Navigate to Maintenance > Order Guides > Items tab

The Prediction Value field allows for more control of each of the prediction methods.

  • Actual and Ideal Usage prediction allows for the entry of a Floor or never falls below level. 

Example:  .5 Cases of 7 oz Steaks is the level I want to never fall below in the store.

  • Reminders method it allows for the entry for the number of orders in between reminders for a product.

Example:  Remind the store every 20 orders to buy 1 box of crayons for kid’s menus.

  • Sales based methods it allows for the entry of the number of dollars of sales per case of an item.

After the Prediction methods and parameters are complete, the stores are ready to start reviewing the predictions and placing their orders based on that feedback.

Placing Orders with Predictions

Navigate to Operations > Order Entry > Enter Order

When the stores go to the Order Entry page for a distributor that is set up for predictive ordering, they will have the option to select the Order Date based on the schedule that has been set up, or, to do an off-cycle order and set their own order and delivery dates.

After that, they can create a new order with Predictions, Clear Predictions, or Order Off Schedule. 

The Predicted amounts are automatically entered into the ordered Quantity (Qty) field, but the stores have the ability to adjust those based on their knowledge of what is happening within their market/store for the next few days.

For example, a store manager should adjust the predicted quantity should they have a large event like catering or holiday coming up.

(star) Note, Decision Logic stores the Predicted Quantity and the Ordered Quantity for future comparisons.

Calculation Example

If the floors are set to zero for 1234 Shrimp and 4567 Chicken Thigh, the calculations used for prediction, using 6 days until next shipment:

Predicted usage = 0.55661 cases/day

Floor = 0

Inventory on hand = 3.28 cases (155 lbs and 8.955 lbs prepped)

No pending orders

6 * 0.55661 - 3.28 = 0.05966 -> 0 predicted since we round the result

The average usage per day for the last 8 weeks for the shrimp was 0.55661 case, which is where we get our predicted usage per day.

For the chicken with 6 days until next shipment

Predicted usage = 0.01827 cases/day

Floor = 0

Inventory on hand = 1.15 cases (40 lbs and 6 lbs prepped)

No pending orders

6 * 0.01827 - 1.15 = -1.04038 -> -1 and we round the result -> to 0

The average usage per day for the last 8 weeks for the chicken was 0.01827 case predicted usage per day.

You will want to review your minimums for these items and adjust them according to your needs.

Table of Contents

For additional questions or assistance please submit a support ticket by visiting the Decision Logic Help Center.

All training videos are located in the /wiki/spaces/CKB/pages/1121779733