Chatbots in retail are increasing in popularity, as retailers seek to automate and to improve the customer experience. The popularity has been driven by the omnipresence of messaging apps. Retail brands are turning to such platforms to help bridge the gap between online and offline experiences.
However, the most effective chatbots need to keep the customer satisfied and create an experience that leads to a return visit. For this innovations in artificial intelligence are key, according to Rasmus Skjoldan, Chief Marketing Officer at Magnolia CMS.
Digital Journal: Why are chatbots an effective way of communicating?
Rasmus Skjoldan: If you look at what has happened on the web in the last 20 years, we’ve all been so fascinated by the instant availability and the richness of information that can be accessed. As humans, we’re still adapting to the fact that knowledge isn’t something that needs to be internalized or kept record of, but instead can be looked almost instantly. We’ve gone from “I don’t know” to “I didn’t have time to look it up”.
While this has been a fantastic evolution in so many ways, it’s also clear that everyone using digital media in a bigger way is kind of going at it alone. Further, when it comes to interacting with brands, a lot of marketers are still trapped in a somewhat traditional mindset centered on broadcasting advertising.
To combat this, a range of technologies try to soften that approach and mimic real human communication. One key stream is personalization and another is chatbots. We see a variety of examples of chatbots starting a conversation but eventually you’re passed on to a real human being. The reason chatbots are successful is that they bring the more human-like communication to digital media. So they’re a sign of a craving for more dialogue and less receiving of broadcasted experiences.
DJ: What makes for personalized customer experience?
Skjoldan: The key here is really understanding the context of the human being on the receiving end. What typically happens is that marketers don’t reach a level of great personalization because as a website and brand, they’re essentially clueless on the visitor’s preferences and situation in that moment. What we see too often is a sort of awkward offering presented to you due to the system not exactly knowing what to put in front of you.
When it does work, it’s invisible. Visitors don’t even notice that they’re getting the right offer and think they’re getting the same one as everyone else. Perhaps they even feel that it is an offering that matches their personal preferences to a tee, but they’re not actively thinking about it.
DJ: How else can online retailers use AI/ML to automate their approach to content creation and curation?
Skjoldan: Content creation and curation are two very different things, so I’d like to break those two down independently.
Creation: When it comes to actual assisted or complete ML-driven content creation, we’re still in the early days, where successful machine learning-provided content is entirely created by the machine. A few examples are sports results, weather and stocks – all areas that a machine can produce content around because of their finite number of variables. When you talk about marketing content in retail, or experience fragments for example, we are still very far off from the machine being able to create with accuracy.
A lot of times, brands are capitalizing on machine learning as an assistant to marketers. While the machine won’t create the content, it will create a lot of stuff around it. For example, if the marketer writes a paragraph, the machine will read that, draft a piece of copy and then suggest images or tags that match it. It may also pull up relevant information from the web as research material. So, marketers are getting instantly informed by the machine about how to further improve their content. This is prevalent to retail as finding the right image assets is key.
Curation: There are examples of retailers who use content from their suppliers. Say a large retailer asks who uses content from their suppliers. They will then ask all of their suppliers to give them content assets. Then, all the marketer on the retailer’s side has to do is curate that content and put it inside their own branded content.
That is where ML can really do a lot in terms of assisting marketers — by scanning all of those materials to ensure they are pre-processed before action is needed on the actual marketers end. The machine can also come up with suggestions about which assets can be put together in a way that makes sense.
