When AI Met Retail: The Story of Predictive Intelligence

Following a two-month hiatus, CEO.digital’s long-running webinar series All Things AI: The Good, the Bad & the Ugly returned with a new theme and a fresh selection of AI use cases to rank.
This time, we set our sights on artificial intelligence in retail and discussed the myriad ways that AI is revolutionising how customers shop. Watch the full webinar for free now.

Once again, webinar host and CEO.digital Managing Director Craig McCartney was joined by chair David Skerrett, who together introduced and set the scene surrounding AI in the retail sector. Paul Winsor then took the webinar’s reins and delved deep into the subject, highlighting the truly transformative nature of AI in retail.

Here’s a summary what we discussed during the webinar. And, of course, you’ll find out how our audience voted on the intriguing use cases our presenters selected – were they good, bad or ugly?


Artificial Intelligence in Retail – Driving Retail Evolution

Once everyone had been introduced, David set the scene and discussed the transformative role of artificial intelligence in retail so far.

As David pointed out, the retail world has come a long way since the industrial revolution. It has evolved rapidly, transforming from the corner-shop where shopkeepers knew every customer’s name into shopping malls that squeezed out smaller stores. Today, it is the shopping malls that are under stress as ecommerce giants continue to alter the traditional retail environment in a digital age through same-day delivery and hyper-personalisation.

It isn’t just the retailers that have evolved. Customers have changed drastically, too, with technological advances fuelling expectation inflation among the masses. Retailers have had to respond to these demands, and they have done so through big data, digital personalisation, omnichannel CX and AI.

Now, the face of retail is a far cry from what it was, powered by AI. For instance, 28% of retailers deployed AI in 2018, compared to just 4% of retailers in 2016. Meanwhile, 41% of the top 100 retailers with $10bn revenue are working with AI. These retailers are leveraging AI against ecommerce personalisation, data analytics, automation, NLP to increase sales, detect fraud, improve CX, automate workflows and provide predictive analytics. In particular, retailers are using AI to handle and process vast amounts of data to compete more efficiently.

It is about utilising AI to take the pain out of the shopping experience and to put the personal, social and anticipation in.

Retailers know how crucial this is to their continued success – retail AI is the hottest category when it comes to acquisitions in 2018/19. It isn’t only the big tech giants in FAAMG group who are jumping on the wagon. McDonald’s, Ultra Beauty and Nike have all acquired AI start-ups in recent years, with McDonald’s paying out $300m for an AI personalisation platform provider called Dynamic Yield.

The reasons for retailers adopting AI are vast and varied, but for Tom Pinkney, VP of Applied Research at eBay, it is because, “AI and ML are driving incremental sales that wouldn’t otherwise have happened.”

As you would expect, eBay is using AI in multiple ways to power the shopping experience. For instance, eBay updated its homepage and added the Interests feature in 2018. This provided the ability for customers to instantly personalise their shopping experience based on passions, hobbies and styles. It is an approach that makes shopping more inspiring by combining machine learning and human curation. Add this feature to eBay’s image search functionality, also powered by AI, and you have a powerful shopping experience that truly understands customers and their wants and needs.


Solving the 3 Main Retail Challenges with AI

It was at this point that Paul Winsor from DataRobot took over the discussion. For Paul, there are three main challenges facing retailers that they must learn to overcome – and AI is helping them. Here are the challenges:

  1. The Empowered Consumer
  2. Product & Supply
  3. Operational Efficiencies

Paul showed us that the modern customer is connected and informed, with evolving needs and habits, that retailers need to understand in order to provide excellent customer service. Meanwhile, retailers must predict product range and price required to meet customer demand and expectations. All this on top of identifying opportunities to decrease costs while enhancing efficiencies to deliver a frictionless customer experience.

You can boil down these three challenges into one core barrier to overcome: how to become customer centric in the modern age.

The use cases to help achieve that include, but are by no means limited to:

  • Predicting next customer state
  • Enhancing customer satisfaction
  • Improving the path to purchase
  • Powering demand, new product and returns forecasting
  • Optimising pricing
  • Selecting new sites
  • Planning staffing and infrastructure

The possibilities for using AI to enhance customer experiences are monumental.


Hyper-Personalisation in the AI Era

To give the audience an idea of the possibilities for artificial intelligence in retail, Paul went on to give an example of DataRobot’s work, selecting its personalisation project with the fifth-largest retailer in the world, Kroger.

Kroger generated $121.16 billion for fiscal year 2019, and a staggering 96% of all customer transactions – or $116.31 billion worth of revenue – are conducted with the Kroger loyalty card. That means that Kroger is able to collect data on 3 billion shopping baskets to reveal what each of the 60 million households that shop with it actually purchase.

Drawing on this immense data, DataRobot is able to help Kroger to deliver a personalised magazine for every customer household, replete with personalised offers that help to up conversions. So far, the My Mag Personalised magazine has delivered over a billion personalised offers to customers.


Enhancing Product & Supply Chains

The personalisation of offers not only helps to give customers what they want – it also helps to streamline supply chains. Paul noted that retailers are generating $1.1 trillion in lost revenue per year due to inefficiencies in product and supply, both because of over- out-of-stocks in retail.

It’s clear what the problem is: retailers don’t fully understand what their customers want or when they want it.

To combat the issue, retailers are moving away from traditional forecasting methodology by injecting machine learning capability into the mix. This means combining internal data such as past sales with external factors, like the weather, holidays, economy, events and more. Then, retailers can pass this through a machine learning algorithm and, after testing, deploy and utilise to make better predictions about the future needs of the business.

One DataRobot client was able to increase their forecasting accuracy by 9%, saving the company over $300 million.


Planning for Staffing Distribution

Finally, Paul turned his eye to operational efficiencies. In retail, employee wages are the second highest cost, which means that short-term spikes in demand can debilitate retailers, especially where costly, short-term contract hires come in.

Another DataRobot client had issues with staffing in a distribution centre, but managed to improve the efficiency with AI. Before AI solutions, the client predicted staffing correctly 67% of the time. With DataRobot, the client managed to improve prediction accuracy by 30% and save an average over $2 million per week during the holiday shopping season by having the right staff at the right time.

Paul had plenty more insights to share about artificial intelligence in retail, and delved into greater detail than we’ve noted here. Learn more and watch the full webinar now.


The Good, the Bad and the Ugly

Following on from our talks, we moved to our recurring segment: The Good, the Bad and the Ugly. We discussed a selection of AI use cases from around the world and opened up the floor to see what our audience thought.

Flu Prediction Technology

Leading US retailer Walgreens has used prescription data and AI to analyse and predict the spread of the flu virus, which is often driven by variable weather patterns.

Every Tuesday, Walgreens updates an interactive map that displays flu activity in the US. It is based on the number of anti-viral prescriptions filled at more than 8,000 Walgreens across the country.

Good 100%
Bad 0%
Ugly 0%

UNIQLO’s Mind Reading Kiosk

Global clothing retailer UNIQLO is pioneering the use of science and AI to create a compelling and distinct in-store retail experience.

Select stores now have AI-powered UMood kiosks that show customers a variety of products and measures their reaction to the colour and style through brain signals, monitored by a small headset. It then recommends products based on each person’s reactions and AI predictions.

example of artificial intelligence in retail
Good 85%
Bad 8%
Ugly 7%

Hyper-Personalisation with My Mag

My Mag Personalised uses AI and data to craft a tailored magazine for customers full of the products, recipes and coupon savings they would love. It is an initiative from US retail giant, Kroger.

The personalised coupons in particular have a significantly high customer conversion rate, which is great for the retailer.

But concerns around data collection could have some customers worried about how their data is used.

Good 100%
Bad 0%
Ugly 0%

AI-Powered Halloween Costume Design

Optics research scientist Janelle Shane has developed an AI solution and algorithm that can pick and design novel Halloween costumes. Shane trained the machine-learning algorithm called textgenrnn to imitate text on a list of 7,182 costumes.

Each round of training expanded the AI’s vocabulary, allowing it to produce more original costume ideas like ‘Ballerina Trump’, ‘Sexy Minecraft Person’, ‘Strawberry Clown’ and ‘Vampire Chick Shark’.

Good 38%
Bad 31%
Ugly 31%

AI Predicts Optimal Staffing Levels

When demand spikes, retailers often have to turn to short-term contract hires to continue operations.

But this is costly and disruptive. AI is helping to predict demand spikes with greater accuracy, thereby allowing retailers to plan staffing levels accordingly across all stores and distribution centres.

The result is an enhanced shopping experience that halts sales losses due to poor waiting times.

Good 93%
Bad 0%
Ugly 7%

Personalisation as Marketing Ploy

HBO’s Westworld returns later this year for a third season, and the network wanted to promote it in true dystopian style. It brought an overly personalised pop-up restaurant to CES 2020 to showcase how creepy AI-data-targeted personalisation can get without privacy safeguards.

Using information gleaned from social media, waiters addressed guests by name, revealed their personal details and even brought the guests personalised menus based on their food preferences.

Good 47%
Bad 33%
Ugly 20%

AI-Driven Inventory Forecasting

Market shifts mean that new products are required to hit market faster than ever before. That means understanding what products need to be where and when to mitigate losses. With predictive analytics, that can be achieved.

These AI solutions allow retailers to improve forecasts for every item/store for the next week, and to factor in returns into predictive stock order. Humans and AI working together to make better predictions ensures retailers have a waste-free inventory.

Good 97%
Bad 0%
Ugly 3%

Read More News Articles

Meet Our Sponsors, DataRobot

The latest edition of the All Things AI webinar series was made possible by our sponsor, DataRobot. DataRobot enables organisations to leverage the transformational power of AI by delivering the world’s only trusted enterprise AI platform, combined with an AI-native strategic success team to help customers rapidly turn data into value. Find out more about DataRobot and to request a free demo of its AI platform now.