How AI can help CPGs define the new normal at the shelf after the pandemic
Time, accuracy and scale: a modern view
The disruptions in consumer demand caused by the Coronavirus pandemic this spring revealed new realities for brands and their retail partners. Panic-buying caused out-of-stocks to soar in sensitive categories, exposing a brittleness in the just-in-time supply chain that trading partners were unprepared to confront.
For CPG brands, the consequence is that shoppers currently are far less likely to find and
purchase the brands and items they prefer. Instead they are much more likely to purchase
the items they can get. This assault on loyalty means executing at the shelf and preparing
for new fact-based conversations about space and assortment have never been more
This paper presents a point of view that AI based machine learning tools have potential to
help brands take positive action to find their way back to the new normal at the shelf.
Shelf presence matters more than ever
As more households have turned to online ordering for grocery products during the
present crisis, the problem of substitutions has come into sharper focus. A recent study by
Acosta, a grocery marketing agency, found that 28% of online grocery shoppers made their
first-ever online orders in March 2020. Eighty-eight percent experienced out-of-stocks,
and of those shoppers, 47% accepted a substitute for half or more of their unavailable
Similarly, recent store tracking report by marketing agency SFW found that 86% of
shoppers who encountered recent OOS had tried a substitute brand. Ominously, 43% of
those shoppers indicated they now prefer the newly-tried brand.
Shelf space changes are coming
Current shortages (and anxieties) have led shoppers to experiment much more freely with
brand alternatives and digital-versus-store shopping behaviors. Collectively, their choices
have exposed certain categories to be under- or over-allocated. Space, adjacency,
assortment – all these fundamental merchandising dimensions are likely to be revisited in
coming months, as trading partners come to terms with the implications.
We can hypothesize that recent substitution behaviors and shifting loyalties will reveal
categories that have been chronically over-assorted. Much of the foundation work of
category planning may need to be refreshed in light of new insights. This sets up a period of
intense competition among brands to defend and re-justify their places on the shelves.
Robots and AI tech generate shelf data
Well before the onset of the pandemic, retailers were already experimenting vigorously
with robotic shelf-scanning devices and camera sensors to capture shelf conditions in real
time and address out of stocks. Best-of-breed solutions use Artificial Intelligence and
machine learning to interpret captured images and deliver rapid and reliable information
on shelf status.
Walmart has led the charge in this area, with a commitment to deploy shelf-inventory
scanning robots in 1,000 stores by year end. The present disruption at retail could help
open the door to more widespread daily use of AI and robotics by retail industry to support
execution. All stakeholders, including brands, will need to reckon with the new torrent of
shelf data that are becoming increasingly available to your partners and/or competition. As
the industry recovers and begins to put its house back in order, brands will want to take
initiative in this area to preserve or even enhance their position on retail shelves.
How can brands lead the way?
There is already strong precedent for this. Snap2Insight has been working with leading
brands including Clorox, and Starbucks, providing them with valuable analytics before and
during the crisis. The solution interprets shelf images collected from retail robots (or
crowdsourcing or merchandisers) and converts them into actionable data using machine
learning. The resultant information can be used monitor on-shelf availability in your
category and detect trends that justify changes in space and facings.
Accurate, timely data will be the key asset brand marketers need to drive better execution
and fact-based category planning. Brands who come to the conversation with clear
analyses of retail conditions and shelf productivity will stand out as value-added partners
as this recovery process proceeds.
Collecting this information store-by-store has been a perennial challenge. Anecdotal
reports from your field merchandisers are unlikely to deliver the quality of data to support
timely decisions or justify which items merit additional space. For best possible impact you
will need to adopt new tools which bring speed, scale, and accuracy.
Defend Your Shelf
In sum, unless brands maintain a continuous clear picture of your product presence at
shelf, they cannot take effective action to create lasting advantages. A winning approach
will use AI shelf analytics to measure both the “what?” and the “so what?” in terms of out-
of-stock and share of shelf.
Snap2Insight recommends brands to explore our ‘AI shelf tracker’ package that analyzes
images from 200 or more stores in three waves across a period of 3-6 weeks, to measure
SKU-level out-of-stocks for your brand and category-level out-of-stocks. Shelf analyses,
including share of shelf and shelf placement are presented in a live dashboard and
PowerPoint. Full results will be available within two weeks after data collection ends.
What will you gain: Overall out-of-stock rate and dollar loss, which can enable your brand
to hold retailer, merchandisers accountable. Rapid discovery and fixes for out-of-stocks.
Better leverage in the discussion to rationalize assortment and fair share of shelf.
If you deliver the right data, even non-captains will earn a seat at the table in the
realignment to come.