Why data science & AI should matter to senior retail executives
It’s no news that 2020 has been a challenging year for the retail sector. The reduction of physical shopping has pushed digital transformation ever higher on the agenda. As we speak about digital transformation, senior leaders are looking for ways to optimise their organisations which brings up associated topics such as data analytics, data science, artificial intelligence, and robotics. But what does all this mean? Why should you and your organisation care?
To optimise an organisation, you are juggling a combination of objectives:
- Make the right decisions
- Improve operations
- Reduce risks
To achieve these objectives, you require information and technology. First, you need to answer these simple questions:
- What happened?
- Why did it happen?
- What will happen?
- What should we do?
- What if?
- When should we do it?
We can only answer these questions objectively using data. Then, you can apply technology to process that data to give you answers and improve processes. This is where Data Analytics, Data Science, and Artificial Intelligence can help. In retail, because of managing physical assets, you often also see investment in Robotics. How can each of these help you?
- Data Analytics and Data Science: provide insights and foresights to optimise business performance with data.
- Artificial Intelligence: improve, augment, or automate a manual process thanks to intelligent algorithms that learn from data.
- Robotics: physical programmable machines that can scale an activity and reduce safety risks for humans. It is often confused with Robotic Process Automation (RPA) which is a different concept tackling automation of digital tasks via a series of rules.
The retail sector is primely positioned to benefit from Data Science & AI. Here are a couple of examples:
- Price optimisation: a revenue growth opportunity
- Stock optimisation: an OPEX efficiency opportunity
- Churn modeling: a revenue protection opportunity
Price optimisation: what is the right price at the right time to maximise returns? You need to learn a relationship between a higher or lower price and the demand seen for that product at that price. This relationship can be further enhanced by considering competitor prices and availability, all of which can be automated to respond dynamically to changing market conditions.
Stock optimisation: what stock and inventory of products should you maintain? Data science projects that link uncertain forecasted demand to products that diminish in value (or must be discarded) over time are important for any large retail business that cares about reducing waste.
Churn modeling: how to reduce the loss of regular customers to competitors? It requires identifying the pattern of behaviour that leads to a lost customer together with an appropriate mitigating action. In a retail setting, this can be tracked by looking at patterns of customer loyalty card usage or credit card usage. In some stores, we may also infer it from face recognition in store.
Once you have identified potential use cases and opportunities to deploy data science and AI, it’s important that you undertake a feasibility study to decide what to prioritise. There are always costs involved that have to be evaluated against potential return on investment. If you are looking at optimising a process, you have three categories:
- Improve an existing process: e.g., help customer support teams identify and classify important support tickets quicker through a better user interface
- Augment a manual process: e.g., help customer support teams to do their jobs faster by pre-populating replies for common questions, freeing them up to focus on delivering a better customer experience
- Automate a manual process: e.g., fully automated customer support for example using chatbot technology
There are several questions you should consider as part of your opportunity discovery phase:
- What problem are you seeking to address or the need you are trying to serve?
- Can you measure improvement in the process performance?
- Does performance improvement generate proportional business value?
- Will AI improve performance compared to the current approach?
- Can you collect additional data to make it better?
- How expensive is data collection, cleaning, and labeling?
- Can you tolerate the system to have an error rate?
- What is the cost of potential errors?
If your organisation is new to Data Science & AI, a recommended starting point is to focus on an existing process you can demonstrate has a worthwhile monetary impact when you improve it using data. Focusing on high feasibility opportunities will allow you to develop confidence, build organisational capacity and trust and reduce the time to seeing a return on investment before venturing into more ambitious, riskier projects.
Now more than ever, management & leadership must adapt, step up and maximise the potential of their workforce, data and technology infrastructure. To do so requires investing in education, change management, new roles, and identifying new business opportunities.