“Data-Driven Thinking” is written by members of the media community and contains fresh ideas on the digital revolution in media.
Today’s column is written by Catherine Leung, managing executive director, head of activation, Hearts & Science.
As ecommerce continues to boom, AI technology promises to revolutionize nearly every aspect of a retailer’s operations. At the same time, products frequently explode overnight thanks to unforeseen TikTok movements or the whims of digital creators – something that’s almost impossible for machines to anticipate.
Never before have technology and data played such a powerful role in driving what people purchase. Yet, the human aspect of tastemaking has never had more influence.
This paradoxical reality is a stark reminder that, even as retailers move to embrace automation, the need for human oversight and decision-making can’t be overstated or overlooked. The key to success for businesses will lie in determining where AI and machine learning will have the most meaningful impact and recognizing what machines can’t do.
AI’s advertising promise
AI can accelerate commerce in many meaningful ways. It can scale any data-driven process, provide predictions based on patterns and make informed recommendations based on a variety of insights and data sources. Below are just four areas where it can have a significant impact for advertisers:
1. Demand dynamics: The right technology can help retailers understand demand dynamics at scale, parse how audiences around the world react to external trends and uncover related behaviors that might not be intuitive. Currently, demand planning is the most widely used machine learning application in the supply chain. But going forward, AI promises to benefit replenishment timing and retailer commitments, too.
2. Supply needs: As the pandemic has taught us, machines are much better than people when it comes to monitoring sales and supply chains. It can anticipate needs before problems arise.
3. Inventory management: Algorithms excel at helping companies monitor what is currently on their shelves, what’s available in the warehouse, what needs to go where and how long it will take to get there. Connecting live inventory to relevant offers and promotions is only possible when you know where your products are in real time. Automation can create symmetry between demand and available products.
4. Paid media: The combination of programmatic ads and behavioral and sentiment data is a potent way for brands to push the right products and offers to the most receptive audiences in moments of intent. Machine learning has the power to drive incrementality through tactics like audience segmentation modeling and dynamic creative optimization. But success requires a solid foundation of inputs (e.g., strong creative and messaging) from which to augment.
Humans at the helm
Despite the potential of AI to transform many areas of business, managing a commerce business still requires skilled experts to mind the store, so to speak. While models are valuable for brands when planning and making decisions, even the most sophisticated retailers need a human to determine where opportunities for refinement – or even innovation – lie.
1. Humans must connect the dots: AI will never be a set-it-and-forget-it solution. The dirty secret of AI-powered ecommerce today is that, while many solutions pitch themselves as one-stop shops, they are actually point solutions that create silos. Connecting these systems is a big lift and requires human judgment.
2. Tastemaking and culture can’t be replicated by machines: It’s dangerous to extrapolate any trend over time. The “new normal” keeps changing, and people are inherently unpredictable. When it comes to shopping, there will always be a very human desire to curate one’s life – and tastemaking isn’t something easily fed into an algorithm.
As brands examine which parts of their business lend themselves to employing software and algorithms and which don’t, it’s crucial to dispel the notion that a single AI solution is possible – or even desirable. When making choices about where to place bets, retailers must assess what can be automated and what requires human oversight. Additionally, they must assess which aspects can be handled internally and which areas will require outside help to orchestrate.
As software promises to bring massive scale and efficiency to ecommerce by unlocking demand, shopping is still fundamentally about brands. Tomorrow’s leading commerce players will be the ones that strike the right balance between automation and human touch.
Follow Hearts & Science (@hearts_science) and AdExchanger (@adexchanger) on Twitter.
