A few weeks ago, we had the pleasure of participating in the Tech Festival, one of London’s best retail tech events, and one of the most talked-about events in digital commerce and the technology driving it. Retailers, tech suppliers, FMCGs, startups, investors, and analysts all joined forces at the event to work out how they will not just survive the digital revolution but nail it.
Retailers and AI: Our take from the event
So many people are talking about the digital revolution, data, and AI, but very few retailers are using their data efficiently and in meaningful ways. Fewer even use it for predictions despite the vast advantages predictive analytics can bring to the (business) table. Despite the big hype, it turns out that only about 5% of businesses use predictive analytics today. Data shows 87% of organizations have low BI and analytics maturity (Gartner 2018).
Predictive analytics is not so simple as you might think.
- Extremely high cost – According to Forbes, the average predictive analytics project is likely to cost millions of dollars and will last months. This is mainly due to amount of manual work of data scientists, which are of incredibly high demand, and require long and extensive effort to recruit and train as well as the time it takes to create and train the machine learning models in order for them to be effective and deliver meaningful results.
- Expertise – Building and maintaining predictive analytics into your organization requires a high degree of unique knowledge to be implemented in house or outsourced by data scientists, build the frameworks, the models, train the models and then hypothesize your business use case, not to mention the time it takes. Recent data shows that a vast 87% of data science projects never make it into production. So many of the organizations that “dare” to implement machine learning applications have shut down the project or understood that they need the help of external consultants or vendors.
- Performance degradation – Even if you’ve managed to tackle the above challenges, predictive analytics has an inherent flaw, every few months existing models will need to be redeveloped, re-trained and adapted (this is known as drifting).
So, in essence, if AI is not within your core business, you’re probably better off to use off-the-shelf solutions.
The holy grail that’s hiding in plain sight
We believe the key to the future lies in your data and your ability to make smart use of this data to improve and better serve your business goals.
While many retailers may fear the digital data revolution or disruption as they call it, we see this as an exciting opportunity to grow, reinvent, and create extraordinary value. We think the future of retail could very well be as bright as ever – that is if retailers would have the solution for rapid-low cost implementation of AI in their day-to-day business operations.
Admire your data!
Dive into it, slice it, dice it, and then you can begin to understand and deduce causality, trends, predictions, and impact.
- Adopt sales and inventory predictions
Sales and other predictions enable retailers to create a highly accurate business process based on a “magical” view of the future – which items are more likely to be sold? dead stocked? Etc. Based on these, it’s a no-brainer to adjust your stock levels, prioritize different items, or even negotiate and change terms with your suppliers.
- Check out payment or credit default prediction
This allows a merchant, a manufacturer or a supplier to predict which customers are going to have issues with their payment and are unlikely to pay on time. Big savings in razor-sharpened’ preventive financial and commercial business execution lies here.
- Implement dynamic pricing
Dynamic pricing essentially allows e-commerce retailers to create different pricing per individual customers based on his/her behavior or patterns. Say I’m a customer looking to book a hotel in Paris, as I’m glancing and finding out more details about the order, or taking the time and not booking. For the next customer – the merchant can adjust the price, such price hiking may also instill a sense of urgency and encourage the customer to take action, purchase the product.
- Examine product-based data and preferences
This is important especially when it comes to delivering superior customer experience. With today’s algorithms we know, to a great degree of certainty, that if a customer had purchased product X, they are also likely to want to purchase product Y. In short, we can recommend additional products based on his purchase behavior, that he/she will most likely purchase. This will of course increase sales and make a happier customer.
Implementing the above recommendations can quickly translate into millions of dollars in immediate value and propel your business forward, especially when aligned with strategic goals.
To sum up, we believe that implementing the steps we’ve covered can help retailers significantly optimize their value chain. This includes working with suppliers through inventory management, eliminating credit default, prioritizing customers that will actually be able to pay all through to implementing dynamic pricing and improving the customer experience by using data analysis.
We’d like to thank Retailweek and World Retail Congress for inviting us to the Tech Festival. This has been a productive and valuable event for us, and we look forward to meeting again in March for the next event!