In a year when the pandemic ground in-person interactions to a halt, it’s not surprising that we’d turn to digital channels to fill the gaps. And when it comes to chatbots, our latest Robots Among Us data saw fashion chatbots jump more than any other type. We’re not just using them more — we’re liking them more.
The jump in comfort for this technology use case does show us some surprising developments that will not only help us make the best use of chatbots in retail settings, but also for how to bring the technology to other scenarios that would benefit from an AI-powered boost.
Better Systems
One of the reasons we’re seeing a bump in chatbot popularity is that the recommendation and language models are getting better. The technology has existed for a while, but as is so often the case with new tech — especially AI-powered, data-driven tech — capability hasn’t quite been at the level necessary for mass adoption.
But this is changing, and rapidly. Chatbots are increasingly capable of nuanced natural language processing, and as the algorithms get better at processing the sense of things as much as the meaning, they have led to better end-user experiences.
With the advances in natural language processing, every single chatbot interaction stands to improve as users don’t get derailed midway through an interaction by a bot that does not understand a command or response.
Similarly, we’re seeing more refined recommendations from the algorithms. More data, in particular more user feedback data, and better data organization means that recommendations are no longer as reliant on broad categories (for instance, all coats).
Instead, they can start to hew much closer to the specific items a person wants based not only on the broad need (cold-weather apparel), but also their individual tastes (from favoring neutral tones and natural fibers to pricing and sizing aligned to past purchasing behavior). This is a huge part of how fashion chatbots have moved from novelty to essential tools in the purchasing process.
Not every experience needs to surface recommendations about a product, but being able to make effective use of data to customize an experience for your end users is crucial to making these new experiences stick. Being able to surface content tailored to their tastes increases the value of these interactions significantly.
Better Reasons
Another important shift we’ve noticed, which relates to the improvements in the technology, is that utility is trumping technology. There may have been initial interest or buzz in fashion chatbots when they launched, but using the technology just for the sake of it had diminishing returns, and we were seeing that in survey results from a year ago.
But when stores closed their doors and digital became the primary venue for retail, the utility of chatbots shot up. Necessity drove this technology in retail, and that’s a trend we see across several technologies and industries.
When it’s technology for technology’s sake, consumers are lukewarm on it. Augmented reality (AR) translation apps, for instance, are a top 20 technology, coming in 19th out of the 70 we asked 1,000 respondents to rank. But AI smart mirrors — also an AR real-world app — don’t have the same cache at 58. That’s due in part to the utility of each. Translation apps offer immediate assistance in specific scenarios. Smart mirrors are maybe fun to play with, but they aren’t a necessity.
From a business perspective, this is a very good thing. Category-leading brands are changing how they’re presenting fashion chatbots and reconsidering how they obtain a return on investment (ROI). By focusing on the utility and not the novelty of the tech, businesses are not just better able to serve a product to a customer in the moment, they’re actually harnessing the feedback loops to learn more and create better experiences overall.
This more utility-focused view provides retailers and customers a better fashion chatbot experience while lasering in on a utility that can have a major impact on other use cases as well.
Better Implementations
Based on our survey results, consumers are pretty universally uncomfortable with anything that reads as too human. Whether it’s a full-on robot or something seemingly pared back, like a chatbot or voice assistant, once it passes the point where we immediately understand it as a robot, we lose interest in it.
For brands that have been working to carve out a special identity via these kinds of interactions, like a mascot or virtual persona unique to their brand, this can be a tough lesson to learn. Customers aren’t interested in your chatbot, they’re interested in any chatbot that helps them accomplish their goals faster. This isn’t new — Ask Jeeves lost out to Google in the early days of online search because Google’s far superior utility trumped any efforts to build loyalty around a personified butler character.
Consumers want capable, fast, and, most importantly, reliable assistance. They don’t want something that feels like it’s trying to trick them into buying something they don’t need or giving up information they don’t want to share. And when we push the identity facet of a bot too far, that’s the perception we create.
Fashion chatbots take customers beyond narrow browse and search functionality to a place that can help them build the confidence that the purchase they’re making is the right one for them. Natural language processing can begin to learn more actively about customers so that the bot isn’t just guiding a customer to an item, it’s learning their tastes and evolving the process to help match them to the items that best suit their needs.
This improves both short- and long-term outcomes for the brand, and it’s easy to see how other use cases that can harness this technology in similar ways can reap similar rewards.
