THANK YOU FOR SUBSCRIBING
Demand Forecasting From the Hashtag World
Sean MacCarthy, Global Head of Analytics and Store Segmentation, Claire's Inc.
Popular social media platforms are now thronging with consumers’ posts about places, activities, people, and products that are receiving a lot of attention, both good and bad. It is a potential gold mine for a retailer. Fast fashion brands like Zara, Uniqlo, and H&M have the ‘wow’ factor of bringing the latest trend on the runway into their stores at a reasonable price and in a matter of weeks. Being on the cusp of a trending interest area is the next big step for the retail business. The possibilities aren’t restricted to a demand adjustment standpoint; they can extend to rising and falling popularities of competitors, propensity relationships between products, and hordes of local data that can be captured by replenishment, forecasting, or third-party systems for the operations group to act on. Social media is a great place to identify local events that have received sudden spikes of attention, and the events themselves may mean a lot to the business depending on the vertical it is in. For a convenience store like Walgreens, a local marathon that is receiving a lot of RSVPs may mean something, and they can stock their water supplies along the stores that are proximate to the route. What interests me is how we can gather knowledge of such happenings and present it in front of the right people in the right manner and at the right time.
Manning the Machines
There is a plethora of tools that use artificial intelligence (AI) and machine learning (ML) algorithms to comb the social media world for keywords and hashtags on Instagram, Twitter, public Facebook, and public SnapChat.
It’s amazing how far we’ve gotten in terms of intellectual curiosity. Nevertheless, the scenario where the output of intelligent systems can be translated directly to an operational input is still evolving. A few years back, during my time at Sears Holdings, I worked with KickFactory in helping to identify and engage potential appliance customers on Twitter. In our test phase one of its detected tweets read: “Need a new refrigerator. Uncle Jeb just died! #redneckfuneral.” The AI was ready to respond to what was obviously a joke. The platform was technically not wrong in detecting the tweet, but it didn’t understand the humor either! KickFactory utilizes a hybrid approach because of this, where the machine is taught certain contextual and textual nuances by humans until a time where it can ‘think’ in a humanized way.
Social media is a great place to identify local events that have received sudden spikes of attention, and the events themselves may mean a lot to the business
Surveys are showing a growing willingness by consumers to engage with AI and Machine Learning platforms, but if you’re not delivering value with them, they’re just toys. Passing the Turing Test is not enough anymore: just like when interacting with an in store associate, a personalized, empathetic interaction that delivers a value, monetary or not, is what consumers are looking for. Consumers are even willing to allow companies more discrete access to their private data that only a few years ago no one would’ve imagined them willing to disclose. Look at the adoption of smart home devices with microphones, fitness trackers, location data, etc… all being shared either socially or with companies because they bring value. Combining this, while of course adhering to any 3rd party platforms policies and government regulations, companies have a great opportunity to not only detect but influence demand signals as they emerge.
Prepping for the Future
Companies need to remain flexible, not only do the social media platforms de jure change, platforms relevant to one company’s customers may be different than another’s. Reading and influencing demand will continue to evolve, and it is an exciting space, but there’s opportunity for companies internally as well. As average employee tenure decreases and the workforce evolves, knowledge of a specific category or market segment becomes less enduring, but this presents an opportunity thru good master data combined with its history for AI and Machine Learning to deliver the insights that used to take years of experience to gain and play a role in influencing strategic decisions. What are the important holidays, product colors, geographic idiosyncrasies, etc… These are questions that AI and Machine Learning can answer and deliver to buying and planning teams, whether they’re in their first week or have been in the business for 20 years.
My background in virtue epistemology (a branch of philosophy) has taught me that humans are creatures of habit, and retail has taught me, so are consumers. While philosophy has taught me to play ‘Sherlock’ around the most significant questions in life, at Claire’s Inc., I play Sherlock around the ever-evolving habits and demands of the consumer. My job is knowing my consumer—a young girl or woman, who may not always be the one with the wallet, but is potentially relying on her parent or friend to get her around or make purchases. When assisted with the growth and developments in technology and social media, it is undeniably a quickly growing world of possibilities and we’re still part of the wave of first explorers.