Warehouse shelves
By Jasneet Kohli | October 13, 2022

Optimizing CPG E-commerce Forecasting on Amazon

Amazon offers a lot of rich data to sellers that can give them a leg up. The best way to leverage it is to “speak Amazon’s language.”

Back in those nostalgic “before times” of late 2019, I first looked at Amazon’s widening influence over consumer packaged goods (CPG) e-commerce in our antuit.ai blog noting: "CPG companies' online channel is rapidly growing … with sales expected to double in five years.” 

I think we can all agree that was a conservative estimate. In less than one month, there would be a 35% spike in e-commerce CPG buyers, in turn forcing Amazon to hedge on its landmark two-day Prime shipping promise, and waiting for the new desk I ordered would become an extended real-world exercise in patience. 

While the worst of the Covid-19 pandemic is hopefully behind us, the effects it brought to the e-commerce landscape are here to stay. The consumer goods industry must continue to adapt to our “new normal” e-commerce economy, in which the rules have been redefined by Amazon and other major players. For Amazon-channel sellers, poor forecast accuracy remains one of the largest challenges, as it can significantly affect CPG’s Amazon sales. Prior to the economic upheavals of the pandemic, leading players typically operated at less than 50% accuracy. Following the closures and lockdowns of early 2020, those numbers cratered even further. 

But Amazon's operational practices complicate the matter.  

For example, multiple sellers may be offering the same item on the Amazon online marketplace, including Amazon itself. Yet, there can only be one "buy box" – otherwise known as the buy button the shopper sees after selecting their preferred size/color/quantity of an item. So, Amazon must determine which seller will get that golden ticket “buy box” (and the added sales that typically come with it). 

Amazon uses a proprietary algorithm for determining which seller gets the buy box for every customer visit. Yet we can assume three influential factors: 

  • The item's price (offered by the seller)

  • The seller's stock availability  

  • The seller's reputation and competency: sellers that stock items at Amazon’s warehouses, listed as “Fulfilled by Amazon” (FBA) are preferred over the “Merchant Fulfilled Network” (MFN) items shipped by the seller.

So, we can see that the better the forecast, the better item availability, the better chance of landing the coveted buy box. But Amazon's operational processes make forecasting difficult for many CPG companies. For example, despite their global network of massive warehouses, Amazon most often cannot keep large amounts of stock on hand for individual products. As a result, reordering intervals are short. For fast-moving items, Amazon can submit a purchase order as frequently as twice a week. Agile fulfillment becomes critical. 

However, Amazon does provide some unique data sets that can be used to quantify the demand drivers and constraints on historical sales. These useful advanced metrics include:

  • Glance views: How many times each item's page has been viewed in a week 

  • Unique visitors: The number of unique visitors who viewed each item's page in a week

  • Total number of customer reviews per item, updated weekly 

  • The average customer review ranking per item, updated weekly

  • The ranking of the item's page unit sale and sales amount compared with other items within the same Amazon category or subcategory 

  • LBB (Lost Buy Box): The number of times that Amazon has lost the buy box to a third-party seller because of the pricing

  • Rep out-of-stocks (OOS): The Number of times that Amazon has lost the buy box because a replenishable item has been OOS 

  • Sellable and unsellable on-hand units: Sellable and unsellable stock per item per week

It is important to remember that Amazon delivers this item information by the Amazon Standard Identification Numbers (ASIN), not specifically to the individual seller's item, as there may be multiple sellers for the same item. However, the information is still very valuable. The number of glance views can reflect the sponsoring of some items during paid promotion programs to bring the item to the top of search lists. Changes in sales ranking can reveal dynamics between the sellers' promotion strategies and other competing products. Another rather unique Amazon metric is Lost Buy Box (LBB), which can be used to estimate or refine third-party pricing strategies as well as factor into promotion planning. 

We’ve observed many omnichannel CPG companies struggle to get ahead of the curve when they don’t fully “speak Amazon’s language.” Yet by leveraging this rich data that Amazon provides, CPGs can dramatically improve forecast accuracy, optimize service levels, and achieve the lion’s share of the buy box—rapidly translating into higher sales and revenue. This is one facet of the demand intelligence solutions antuit.ai delivers for our CPG customers.


You May Also Be Interested In This Insight:

Balancing Retail Orders in Times of Shortage Can Be as Simple As An Extension

Manufacturing, Retail, Warehouse and Distribution, Automation, Best Practices,
Jasneet Kohli
Jasneet Kohli

Jasneet Kohli is head of antuit.ai’s Solution Consulting for Zebra and has more than 15 years of experience generating value for consumer products companies and retailers. He has held various leadership roles across business operations, solution strategy, and customer success. Previously, he served as head of procurement and logistics for Abbott in Singapore and Managing Consultant for IBM.

Zebra Developer Blog
Zebra Developer Blog

Are you a Zebra Developer? Find more technical discussions on our Developer Portal blog.

Zebra Story Hub
Zebra Story Hub

Looking for more expert insights? Visit the Zebra Story Hub for more interviews, news, and industry trend analysis.

Search the Blog
Search the Blog

Use the below link to search all of our blog posts.