The original 1-800-Flowers chatbot demonstrates that this company is no laggard. They launched their Facebook Messenger chatbot in April 2016 at the F8 conference. What better press than Mark Zuckerberg announcing “To order from 1-800-Flowers, you never have to call 1-800-Flowers again.”
Well things have moved on and we see a new version of 1-800-Flowers chatbot powered by IBM Watson. However, we did want to go back to see the original experience before we checked out the new one. At 1080bots we keep track of Ecommerce Chatbots to help our clients create the best conversational commerce channels.
The opening dialog asks whether you want to place an order or talk to someone. “Talk to someone” is the legacy of running a phone based service reinforced by the name “1-800-Flowers”. The bot then immediately asks the delivery address. Understandable because they need to work out if they can deliver there, but a raw and intrusive user experience.
In the case of purchasing flowers, customers are likely to be happy or sad, so both sensitivity and inspiration might be warranted here. I am sure that if you called a 1-800 human they might be a little more polite. Think of the bot as a frontline customer service agent.
The bot does resolve the address and finds the zipcode which is useful It then immediately provided a selection of options. Instead of pressing the ‘Thank You Flowers’ button, I typed “Large Thanks Flower Bouquet” – more text than the option. The bot was unable to pick up even the keywords and repeated the carousel. I selected the Thank You Flowers arrangement from the carousel.
Next refinement was another carousel and ‘Foliage Fall Pumpkin Hayride Arrangement’, so I typed ‘Pumpkin’ – less text than the button. Once again the bot failed to pick out my keyword.
I had a few more tries typing and the bot was not responding well. Feeling guilty the experience was so fragile, I decided to follow the 1-800-Flowers Chatbot by using the buttons. It took me through the arrangements and allowed me to select a gift I was happy with. It then had me provide details for delivery including a message.
Once the details were in, I was not ready to purchase so I typed ‘Cancel’ during one of the input dialogs. The bot was smart enough to try to find ‘Cancel’ on a map and came up with a town in France. That was an unexpected and funky experience. At 1080bots we always recommend that there are certain universally recognized words like ‘Stop’, ‘Cancel’, ‘Start Again’ and ‘Hello’. Those words should always be recognized at any point in the dialog and have a predictable and useful action.
Next morning, I did get a message from Stacy, a human who realized that the conversation stalled – the chatbot equivalent of dropoff.
The 1-800-Flowers chatbot on Facebook was clearly a first generation bot that put the company on the radar as being leading edge. There are a few take-aways from this experience:
1-800-Flowers are not going to get left behind, taking all the learnings from this first iteration, they have created a whole new experience on their site itself. We also had a conversation with Gywn, the 1-800-Flowers Chatbot.
If you’re thinking of building a chatbot to take orders or to provide customer service for your e-commerce channel, just email us at firstname.lastname@example.org