We’ve been hearing a lot from IBM about the Watson Platform for Conversational Commerce, so we were excited to see how Gwyn, the 1-800-Flowers chatbot would perform.
Hi, I’m Gwyn: Your Virtual Gift Concierge
Gywn appears on the 1-800-Flowers website and worked well on the mobile version of the site where we tested it. The quick input of delivery zip code, occasion, and delivery date is a great option, easy to do on a responsive UI but quite tricky in a messaging app like Facebook Messenger or Kik.
Gywn asked what we were looking for and provided an example, “I am looking for some flowers for my wife”. We love giving customers examples, it makes for a nicely intuitive interface, this prompt did double duty because it was also trying to engage in natural language conversation, perhaps to gain a richer set of terms than a site search query. The words “I am looking for some” are probably unnecessary in this dialog but serve to make the user experience clearer.
We opened our dialog with “Large Thanks Flower Bouquet”. The Watson chatbot did really well and returned a bright and inspirational carousel of Thank You bouquets. Among them was an item named “Sunset Passion”.
We tried typing in the word “Sunset” but Gwyn wasn’t able to understand. This made us wonder if the 1-800-Flowers product catalog has only been attributed with a set of curated terms corresponding to possible user intents.
At the foot of the chat window was a set of buttons, we clicked on “Products” and were shown a carousel of the items we had seen so far, a handy feature as we browse.
Real Men Don’t Want Flowers
We decided to try a different tack – what might the bot select if we were buying for a man? The bot suggests flowers for 18 and over. We tried again “Man’s birthday small arrangement”. Gywn is not keen to send flowers to a man, so she offered up some meat.
We moved on to try to describe what a florist might suggest for a man. We entered “Flowers dark colors lots of greenery” hoping to give the 1-800-Flowers chatbot enough words to define our intent. The result looks very keyword-ish as Gywn returns with “Dark chocolate is so delicious and has health benefits”.
We tried something simpler “Brown and Green Bouquet” Gywn came back with a very bright set of options.
Search 101: “Brown Bouquet”
The selection of colors proved difficult for Gywn (reminding us a little of the Sephora bot and the red lips). We entered “Flowers, Blue Bouquet” and the chatbot returned white flowers. Maybe Gywn was trained against a floral encyclopedia which seems sensible, so we added a flower we know to be blue the “Cornflower”. Bingo! We see a white and blue bouquet on the carousel.
We kept on the trail of blue and white until the 1-800-flowers chatbot realized we did not select flowers and offered us chocolates. This was a great move because the 1-800-Flowers chatbot is getting closer to detecting ‘intent’ or in this case, lack of intent.
What We Learned
Some thoughts on where this chatbot could be improved to help customers shop with the minimum of friction:
- Chatbots should be at least as good as site search – continually analyzing all products including names and descriptions to make sure that it can find products by name
- Fully understand the context of the product – in the case of flowers, colors, flower names and characteristics should be part of the vocabulary
If you’re thinking of building a chatbot to help your customers, drop us a line at email@example.com or fill out the form below and we’ll get right back to you.