‘Today, retailers can harness the power of artificial intelligence in practical ways’

The term ‘artificial intelligence’ can conjure up images of Terminators or HAL 9000 in 2001: A Space Odyssey. In reality, this kind of artificial general intelligence is still a fiction. 

However, retailers today can harness the power of applied artificial intelligence (AI) in more practical ways — leading to massive gains in efficiency, better decision-making and a more personalised customer experience, as well as significant uplifts in financial performance.

Ultimately, retailers’ most prized asset is the data that emits from across the business. AI is the key to unlocking its value. 

In department stores, AI enables an omnichannel experience through chatbots and visual search. In fashion, it helps analyse store receipts and returns to match inventory with demand. 

In big-box formats, AI monitors and manages stocks and in convenience stores it enables customers to pay without queuing. Online, it supports virtual assistants that can resolve customer complaints in seconds. 

And in almost all environments, it can derive optimal SKU-level prices, based on hundreds of experiments across the network. Most of these examples have the potential to be rolled out at scale.

AI’s potential in retail

The prize for AI adoption in retail is about £0.65trn ($0.8trn) of value creation annually. Certainly, across industries, leaders get disproportionate impact from their investments, outperforming their peers by about 8 EBIT percentage points of margin. 

Two areas where leading retailers are most focused today:

1. Taking personalisation to the next level. Targeted messaging and experiences are nothing new, but retailers that invest in engineering and in connecting their data sets can obtain a 360-degree view and truly understand customer-lifetime value. Through AI-driven insights, they can predict behaviours, visits, buying patterns and broader conversations that they could, or should, be having. 
2. The AI-powered store. AI can now enable retailers to optimise their operations, prioritising tasks in real time, for example in response to local environmental factors (traffic, weather, etc). 

AI-based anomaly detection can scan activity to detect unexpected changes and find root causes such as accidental delisting of an important item, a change in promotion strategy or competitive activity.

The next stage will be leveraging the internet of things, including smart cameras and wearable technology, to create digital twins of in-store operations. These can be used as test beds for new layouts or to assess the impact of external factors such as disruptions in supply chains.

Getting started and scaling

Exactly 57% of companies have invested in AI to some extent, but only 24% succeed in deploying AI at scale. Through our work across thousands of engagements, we see three common traits associated with a successful approach:

First, a business-led, value-backed strategy. Leading adopters don’t just behave differently, they think differently. They focus on being truly AI-enabled, capturing and leveraging data at every opportunity. They move from a focus on earnings to impact on multiples. And they shift from buying off-the-shelf AI solutions in the short term to investing for long-term value.

Second, ramping up capabilities. To go from deploying 10 models to hundreds requires talent (creating multi-disciplinary teams), an integrated data strategy, fast and resilient technology and an agile approach to delivery that accelerates development, testing and deployment. 

Adopting practices such as machine-learning operations can lead to a dramatic reduction in the cost of experimentation, enabling value capture at an ever-faster pace. At one organisation, deployment times shrunk from 12 weeks to three. Through these approaches, AI’s impact is growing, with 27% of companies saying at least 5% of EBIT is due to AI.

“AI offers exciting potential across the business”

Finally, retailers need a coordinated execution and change management plan. Culture is one of the most common roadblocks here. Taking managers and frontline workers on the journey, involving them in the development process, leveraging explainable AI to make outputs understandable and upskilling them will improve adoption rates.  

AI offers exciting potential across the business. However, for those seeking inspiration on its criticality, we concur with one chief executive we have worked with: “If you haven’t got good AI,” he said, “in a few years’ time you won’t be on the page.”

• Anita Balchandani is senior partner at McKinsey & Company and Louise Herring is partner at QuantumBlack, AI by McKinsey