How to Optimize Inventory with Demand Forecasting

In this article:

Why Demand Forecasting Matters

Accurate demand forecasting is a must-solve problem for brands looking to improve their margins. Too much inventory on the shelves? Storage costs eat away at the bottom line. Not enough products in stock? Lost sales pile up.  Add in the complexity of solving these problems across multiple warehouses and you're looking at hours of time & wasted opportunity just getting to the insights.  Enter: Demand Forecasting across multiple warehouses.

Cart's Demand Forecasting algorithm ingests your sales and marketing data to create daily, weekly and monthly forecasts for each SKU across each of your fulfillment centers. Equipped with these forecasts, you can dynamically adjust marketing spend and inventory orders to optimize your margins. By default, the algorithm will show a 60-day forecast for each SKU and calculate the cost or revenue impact for too much or too little stock. A buffer of 60-90 days stock on hand is a common best practice in e-commerce, proven by Amazon’s IPI scoring methodology and shared by ecommerce experts. This default setting gives you the confidence to make the best decision possible on where to invest your cash.


Inventory Analytics Overview

You can find the Attract Overview dashboard by clicking here if you are logged into Cart Console, or by navigating to Unified Analytics > Fulfill > Optimize Inventory:


How to Use Demand Forecasting to Optimize Your Inventory

Utilize the "Show Instructions" Button

Within Optimize Inventory there are instructions you can always click on if you want reminders on how to utilize your demand forecasting tool.

Forecast Confidence Global Filter provides full transparency on the confidence level for all of data science models powering your demand forecasts.  This gives brands the ability to trust how confident they can be if they may stock out or have too much excess inventory over the next 60 days.

The top row of Optimize Inventory has a new histogram with the number of stock alerts per confidence band to help brands better understand the state of their inventory. It also features a more intuitive confidence level filter via the low/medium/high designations rather than percentage-based. (See left chart called "Forecast Confidence").  

Additionally, brands can view their stock alert volume by fulfillment center to more easily hone in on imbalances. (See far right chart called "Forecasted Stock Status by Fulfillment Center")


Fulfillment Center Level Forecasts & Filters

Sku-level demand forecasts are now also available by fulfillment center to help avoid excesses and stock outs for customers served by each fulfillment center. By filtering to a specific warehouse, brands can view the largest projected excesses and losses to begin triaging imbalanced inventory for skus in that location.


Stock Statuses

Our proprietary Demand Forecasting algorithm generates sales forecasts for each of your SKUs and projects stock surplus or shortage based on your inventory levels at each of your fulfillment centers. SKUs that have an imbalance between demand and inventory are labeled with a stock warning:

  • At Stock: SKUs with inventory levels that meet projected demand over the next 60 days.
  • Excess Stock: SKUs with more inventory in stock than projected demand over the next 60 days.
  • Low Stock: SKUs projected to experience stock outs in the next 60 days based on demand.
  • Out of Stock: SKUs currently out of stock.

The SKU summary table shows the distribution of these labels across your entire catalog. Clicking on these will filter the bottom portion of the dashboard to only SKUs with that label:

For SKUs that are out of stock or low on stock, you may consider creating a purchase order or shifting marketing dollars elsewhere. For SKUs with excess stock, allocating more marketing spend could help reduce storage costs from excess inventory.


Identifying SKUs for Optimization

The Demand Forecast Details by Sku Table shows a detailed view of the demand forecast and stock status details for each SKU.  You can also view projected stock statuses of skus across your fulfillment centers for deeper granularity and forecasting across multiple warehouses.

Based on the forecasted orders per day and inventory on hand, the table shows a prediction for when stock will run out (Days Stock On Hand). The Projected Excess or Loss column shows the number of excess units on hand for SKUs with excess stock, or the number of projected lost sales from stock outs for low or out of stock SKUs.

How do you use this in practice? Generally, you should order more inventory for SKUs with the highest projected loss if possible and reallocate marketing dollars from SKUs with low/out of stock warnings to SKUs with excess stock.

The below image shows your SKUs sorted by high projected excess and loss so you can easily identify where to focus your attention for ordering more inventory and shifting marketing spend. You can click on the "Show Instructions" button above the Demand Forecast by SKU for a reminder on how to use the dashboard at any point.