When the 1080bots team found this H&M Chatbot on the Kik Messenger platform, we were eager to try it out. At 1080bots we are constantly seeking the best chatbot experiences.
The opening dialog was designed to help the chatbot get a better picture of me as a customer. It asked me if I wanted to see Men’s or Women’s clothing. I cut to the chase and asked it directly for “men’s underwear”. Clearly on a mission, the bot understood I was male and chose to ignore the rest of the request. This was a shame because I really needed to buy some underwear.
We got past the initial hiccup and the bot asked me my age group with buttons to simplify my response. A series of questions followed that showed my fashion style. Only button or similar text responses were accepted. The bot did not have any semantics around the image content, so ignored me when I talked about the images shown. Looking at responses, the H&M chatbot has been given a style of speech that identifies with its target demographic.
Departing from the buttons when asked me my style preference, I said ‘Formal’ and it politely reminded me to stick to the button choices. Given a profile of my style it was ready to make recommendations. It asked me for an item, so I went back to my original request for underwear. Once again the bot was not responsive and provided more suggestions.
Thwarted in my attempts to get some underwear, I decided I’d cover up with a pair of jeans. The bot easily suggested an outfit and provided a total price. Faced with the fact I’d need to go commando in this outfit I reluctantly said it looked great.
Once I’d got comfortable without underwear, it did ask me if I wanted to share the outfit with my friends. Check in the box on social shopping, but I’d like to see the metrics and demographics of the sharing. Deep into this conversation, I was invited to buy any of the items shown. The bot seamlessly provided a link and Kik opened the product page, with that item on sale.
Overall not a bad experience – maybe designed for a younger male who says things like “Coolio” and shuns undergarments. The bot made good use of buttons and images, built a good knowledge of me as a customer and did not take too long to get me to a product page.
In an ideal world the bot would have access to the full product catalog and provide additional semantics on products and better search. I am sure this is coming in the future and in the meantime, H&M is learning a lot about how its customers are going to use chatbots in their purchasing cycle.
If you’re thinking of developing a chatbot to better understand your customers, get feedback on your products or drive a purchase, then drop us a line at firstname.lastname@example.org.