How DataRobot Supports Santa’s Workshop to Optimize Operations and Improve Service
We recently concluded our AI Experience roadshow with a special regional session in Lapland. We have been lucky to hear from an impressive lineup of generous customer speakers globally. Lapland was no exception. One such customer is the lesser known and famously discreet Santa’s Workshop, a global private philanthropic conglomerate, with vast operations in manufacturing, consumer packaged goods, and logistics.
Their CEO, Saint Nicholas, and the wider executive team shared insights into how they’ve used DataRobot to transform their centuries-old operation. It has been a challenging decade for the Workshop. The global population has grown to 7.7 billion, and with their core market of children growing most rapidly, it has been difficult to keep up. Moreover, current generations are more demanding. They expect more real-time customer service and an ever-broader selection of products. Each season there are usually a few unexpected super popular toys, making it hard to predict stock. Artificial intelligence is now at the heart of Santa’s business, enabling them to overcome these challenges by optimizing operations and improving service.
The head of Data Science, Alabaster Snowball, had spent centuries honing his market-leading magical predictive models. However, it takes months of work to keep the models current, gain compliance approval, and deploy them into production. Automation allowed him to boost the productivity of his data team, providing them with more sophisticated and targeted models in a fraction of the time.
Alabaster Snowball also spoke passionately about the power of democratizing AI, “My team was overstretched and under resourced. With DataRobot, we’ve been able to get a wider range of elves involved in AI projects.” Wunorse Openslae, a Sleigh Engineer and Process Improvement Analyst added, “I’ve never written a line of code before, but now I’ve built my own AI to proactively get our sleighs up and running in time.”
The boost in data science productivity and involvement from the wider business has enabled Santa’s Workshop to apply AI to a wide range of previously untouched use cases. These include:
- “Naughty” or “Nice” determination – Determining who has been “naughty” and who has been “nice” is at the heart of the Santa operation, helping the organization to streamline present distribution. Only the deserving are rewarded.
- False positive reduction – Some pretend to be asleep when they’re not and hide their pouts when Nicholas is in town. A huge number of potentially “naughty” cases get flagged for review, but most are false positives which can be cleared using AI.
- Anomaly detection – People have become increasingly sophisticated in their attempts to deceive. Anomaly detection helps the team identify new forms of naughtiness, even when they’re disguised.
- Predictive maintenance – The workshop is one of the largest and most sophisticated assembly lines in the world, able to produce any toy. Monitoring IoT sensor data, vibrations, and sounds generated by the vast array of equipment allows faults to be identified, so that they can make proactive repairs.
- Overtime forecasting – In the cold winter months, elves frequently take sick leave when the need for workers is higher than ever. The Workshop was able to cut the costs of overtime through better forecasting.
- Elf retention – Preparations for the holiday season require long hours and often grueling work. Elf Resources used AI to better monitor the well-being of their elves and proactively identify those who might need more support. The team plans to tackle Elf and Safety use cases next.
- Predicting delivery delays – Unbeknownst to many, Saint Nicholas is no longer able to cover all the deliveries himself and relies on a vast network of global third party distributors. This brings a risk of delayed deliveries or compromising on the brand promise associated with the more personalized service he used to provide. AI is used to predict problem deliveries and allows for interventions.
- Letter classification – Millions of letters (and increasingly, emails and social media messages) are sent by children each year to the Workshop. AI is used to quickly categorize and prioritize these requests. Connecting DataRobot with their in-house RPA tool, the Workshop has automated the majority of the end-to-end process of replying.
- Churn prediction – Every year, many loyal customers are lost as children stop believing in Saint Nicholas and his charity. This can be prompted by various factors, particularly bad Christmas experiences, like receiving the wrong toy the year before. Some even trick their younger siblings into thinking Santa’s Workshop is a fraud. In any case, most people become more demanding and cynical with age. AI helps determine those most at risk of being lost so that the retention team can make special interventions to highlight the magic of the holiday season.
- Global and personalized toy demand forecasting – Correctly predicting which toys are needed and where to send them is very challenging. Demand varies significantly over time and location. The team were inspired by the work of the likes of Kroger and now use Automated Time Series to solve this problem.
AI Trust and Ethics
Saint Nicholas has had a long and diverse career. Very little is known about his activities before becoming CEO of Santa’s Workshop, though he was rumored to have been a bishop in the ancient Greek city of Myra. His habit of secret gift-giving quickly became legendary. This gives Nicholas a deep appreciation for ethics.
Nicholas summarized his views on AI trust saying, “We don’t believe in black box models – I need to be able to justify to any child who was naughty why they didn’t get what they wanted and help them be nicer next year.” Here the human-friendly insights of DataRobot’s analytics proved invaluable. They can understand what behaviors are driving the AI’s decision-making and their relative importance (e.g., the year summary, charity donations, grumpiness, joy, bedtime etc.) This also enables them to remove features that might cause unfair bias, (e.g., “zipcode”). Alabaster Snowball and his team also were able to automatically generate robust guidance documentation to explain the AI models they built, gaining rapid endorsement from the compliance team.
They have also generated their own AI Ethics Guidelines to build consensus across the organization. Deploying AI is about more than just accuracy. Under the executive sponsorship of Saint Nicholas, they deemed it essential to make sure their use of AI reflected the values and brand of Santa’s Workshop.
DataRobot wishes readers worldwide a very happy holiday season. We’re confident that with the power of AI, companies will be able to deliver better services and delight their customers over these busy weeks. We hope the challenges and AI use cases presented above will resonate across all industries and help inspire initiatives in the year ahead.