Opportunities and Challenges – how AI is shaping the future of convenience retailing

Over the last few months AI packaged under the name ChatGPT has gone viral. It is, in fact, the fastest-growing consumer internet app ever, according to analysts (https://www.theguardian.com/technology/2023/feb/02/chatgpt-100-million-users-open-ai-fastest-growing-app)  – reaching 100million users in just two months. To put that into perspective it took TikTok 9 months to achieve the same figure.

The AI chatbot has multiple functions including answering questions in a human-like way, composing essays, describing art in great detail, creating AI art prompts and can even engage in philosophical conversations.

With AI now becoming a part of daily life this opens up a wealth of opportunities for businesses, but also a set of new challenges.

To showcase the capabilities of AI, I asked ChatGPT to come up with a song that summarised the key points in each section.

To first understand how AI can revolutionise convenience retailing, we must first understand the current market landscape.

I Will Survive – Gloria Gaynor

“…the lyrics don’t directly reference AI and the convenience retail industry, the message is of overcoming challenges and coming out stronger on the other side…”

Life on the forecourt used to be simple. According to KMPG 90% of customers would visit a site to refuel their vehicle. Fast forward to 2029 and only 20% of site visits will be for fuel. KPMG predict that the customer demand for convenience will drive product and service offerings. They predict a significant boost in retail food and beverage, particularly high-end restaurants and coffee shops; a rise in adjacent services such as co-working spaces and children’s play areas and an increased focus on EV.

The predictions will come as little surprise to fuel and convenience retailers. EV and hydrogen cars are gaining popularity, governments are legislating to support the transition away from fossil fuels and forecourts all over the world are evolving with an increased focus on convenience.

Beyond the forecourt the world has never been as volatile as it is right now. COVID-19 and the war in Ukraine have created volatility and economic uncertainty. Combine this with wavering oil demand, increased EV adoption, the rise of carbon neutral fuels and consolidation of fuel retailers and the landscape can look a little crazy.

The Times They Are A-Changin’ – Bob Dylan

“…society is constantly evolving and change is inevitable…it reflects the idea that AI has the potential to revolutionise the convenience retail industry, but ethical and societal implications must be addressed.”

Chat GPT has brought AI to the finger-tips of regular people and that is some achievement. However it is not without its controversies or limitations. OpenAI scraped the internet (yes that includes your crazy tweets) to train the Large Language Models in ChatGPT. The technology can remember what a user said earlier in the conversation, allows users to provide follow-up corrections and is trained to delete inappropriate requests. However it can occasionally generate incorrect information, produce harmful instructions or biased content and it has a limited knowledge of the world and events after 2021 unless these are specifically fed into the system.

One of the key issues we are facing with technologies like deep learning and, to a certain degree, generative AI (like ChatGPT) is that we can’t explain how it came up with the solution it did.

From a business perspective being able to explain how the AI has come to a solution is essential in the change management process. By building AI models on smaller data sets this allows us to improve the decision making of the AI and enhances the autonomy in AI systems. Here is an example of how AI based fuel pricing works:

In this example the AI is pricing the red line and the blue and yellow lines are competitors. On the first day the AI chose to compete with the yellow competitor and then shifted focus to the blue competitor the following day. From this test we know that it increased by 4 CpG while keeping the market share, but how did it achieve this?

In order to explain the outcome we need a fully connected model where you can fully trace the semantics behind the decision making.

In this example of an AI model from PriceCast Fuel you can calculate the price elasticity of the supply and demand on station product level down to the minute. Like with generative AI, the AI is trained on samples, however these samples are engineered to keep their semantics such that the end user can gain an understanding of the reasoning behind the decision-making.

In this example a station has three competitors, but which one are they actually competing with? This changes throughout the day. It can come down to location, offers and other facilities such as restaurants or schools nearby. On Monday competitor one (orange) is the main competitor whilst on Tuesday Competitor 2 (green) is the main competitor in the morning and Competitor 3 (yellow) in the afternoon.

When the AI can start to explain how and why it is acting the way it is, it becomes more powerful.

Revolution – The Beatles

“The song speaks to the idea of change and a new way of doing things, which is relevant to the potential of AI in the retail industry for implementing AI in their businesses.”

We know that we will need AI and we also know that first movers have the possibility of winner takes most, so how do we get real about implementing AI in our business?

Start with a clear focus on how to operationalise AI. With PriceCast Fuel the AI is capable of deciding on the individual prices of products, removing the responsibility and tedious tactical work of having pricing analysts set up various rules and defining distance to competitors.

There is no simple way to implement AI in an organisation. AI is a powerful tool capable of seeing multi-dimensional correlations. Having selected the right AI model we get superhuman capabilities and this can be hard to implement. Change management is critical for us when introducing PriceCast Fuel. We often start migrating the current pricing set-up as it stands and involve pricing analysts in the process.

Start small and be focused. Measure the effect of the AI and make sure it fulfils its purpose of delivering a compelling customer experience and results in the expected business outcome. With PriceCast Fuel, we do this by constraining the AI around business rules so that the pricing analysts can control the introduction of AI into the pricing step-by-step – allowing them to approve any price suggestions within familiar KPI’s.

If you want to take the fast track you need to make your data actionable – turning it into insights which can drive decisions for you.

By leveraging AI you can extract footfall data from your regular IP cameras. This allows you to trace the customer’s movements throughout the site. Is the person plugging in their EV car also going into the shop as well? Where in the shop? Is there dwell time anywhere?

Suddenly your video feed has become actionable data from which you, or another system can take decisions.

AI could also be used to create store planograms or even to price products in store based on inventory and ensure benchmark products are priced correctly.

Most importantly, AI can help you understand how, when and why customers engage with your business. Social media likes aren’t enough anymore – you need to understand the intention of each and every customer on your forecourt. When you know that, you can transform this knowledge into business actions, which in turn will help deliver a compelling customer experience.

By knowing your customers it allows you to build hyper-personalised relationships with them that, in the end, foster loyalty. Moving away from general offers and offers customised by certain characteristics toward unique, individualised offers tuned to a person’s specific behaviours and preferences and delivered at the optimal time when the customer is receptive and ready to engage.

KEY TAKEAWAYS

• A Gloria Gaynor said “ I will survive” but to do so you must view volatility as an opportunity.

• The time’s are a-changin’ – AI is already shaping the future.

• Revolution – what can AI accomplish for you by making your data actionable?