Q&A: 1-800-Flowers CIO Arnie Leap on a year of shopping with 'GWYN'

What the e-retailer has learned about conversational commerce from its chatbot powered by IBM's Watson.

A year ago, 1-800-Flowers introduced a whole new way to buy flowers online, with the help of the brand’s first digital concierge, a chatbot named GYWN (Gifts When You Need).

Powered by IBM’s artificial intelligence platform Watson, GYWN simplified the search for the perfect arrangement with conversational commerce. For the first time last April, shoppers could talk to GYWN to get flower recommendations through Facebook Messenger and with Amazon Alexa.  

GWYN was one of the first branded bots in a marketplace timid to create their own. At the time, few brands had invested in AI-powered customer service. In the last six months or so, a slew of brands have introduced their own bots, demonstrating increasingly creative ways to use the technology on multiple platforms. Johnnie Walker offers a guided whiskey tour through Amazon Alexa, Taco Bell delivers tacos on Slack and H&M builds outfits on Kik.  

GWYN too has become more sophisticated, learning from her year on the job. Today, GYWN can make more precise suggestions than she could a year ago, drawing from more than 7,000 products across 1-800-Flowers.com Inc.’s nine brand sites—seven times more products than she could at launch. And instead of delivering answers to questions one at a time, conversations are saved and shoppers can edit a list of keywords GWYN can understand for the next visit.

The bot has grown a lot in the last year, but Arnie Leap, chief information officer at 1-800-Flowers.com, says GYWN still has a lot to learn. Campaign US spoke with Leap about the challenges of being at the forefront of bot technology, the lessons learned and the future of conversational commerce.

How did the bot perform in the past year?
It's been incredibly successful. We found that 80 percent of the customers would come back and use GWYN again. We don't divulge our user activity, but we’re very happy with the volume so far and how people are interacting with the bot. On average, the bot asks about five questions, and it's a two-minute engagement. It understands the intentions of the customer in an efficient manner and makes it a positive experience.

What has the company learned from its first year with GYWN?
One of the early learnings we found with our customer base was they weren't quite sure whether they were speaking to a bot or to a human. We wanted to correct that right away. In order to be a trusted gift advisor, customers need to be able to relate to GWYN. If you were communicating with Gwyn, we wanted to make sure that you understood that she's a bot underneath, not a human. We made it clear by having the bot ask if you want to speak to a customer service representative.

Also, early on, with the way we merchandised a couple of our products and attributes, GWYN had a challenge with recognizing the word chocolate. Whether or not it was a chocolate cookie or a piece of chocolate candy. We had to do some work there as far as attribution, but we were able to get past that pretty quickly.

How did it help the brand to be one of the first to introduce a chatbot?  
We're already solving the commercialization of it, where others are just getting started. So we're in our fourth cycle of decision-making and design-making, where others are only getting started now. This puts us in a very different position and allows us to have more choices and options. It allows us to pivot where we don't believe the market is going versus where it is going.

How will GWYN continue to evolve?
We're very excited about what we're going to see in the next 12 to 24 months. We are introducing the ability to stay within GWYN in order to purchase the actual products that she recommends.  Currently, once a recommendation is made for a product, the product is placed in the online cart and it separates yourself from GWYN. What we're doing is we're adding the appropriate functionality to allow you to order that product directly from the GWYN experience.

What is the end goal?
My end goal is to sit under a palm tree sipping Mai Tais…

But really, this helps us to not be limited to one edge device that a user can engage GWYN with. That’s an incredibly important aspect of where we're going in the future. It's so important to encapsulate GWYN’s capabilities in such a way that it can easily be migrated. So moving to the Android platform, or moving to the iOS platform in native form allows us to continue to promote GWYN’s personality and capabilities across any of the devices that come out or introduce as a platform. GWYN will live beyond just websites; it will live on mobile platforms; it will start to permeate through and become that nice assistant for all gifting needs and live on all devices. There's some exciting things that are going on with Google Home, so we're paying very close attention to that as we're moving forward.

Is this where the entire market is going?
In the retail and the e-commerce industry, we've sort of gone from web, to tablet, to phone, mobile device. There is going to be a migration to other devices, especially on mobile phones. I think it's going to come full circle, meaning with calling. Back in the 1980s, 1-800-Flowers decided that toll-free ordering of flowers on a national level was something that could be accomplished using a phone number. And the most important aspect of a phone number is just one person would call another person and place an order, and that would be done through a conversation. We believe the conversational commerce is gonna continue to transform itself where there'll be less and less typing, and more and more talking.

Do you see other companies keeping up with GWYN?
Well, we've seen quite a few bot companies pop up on the west coast in the last six months. What used to be five or six, is now about three or four dozen little startups that are trying to figure out how to best utilize chat bot and chat capabilities in conversational commerce. And the big guys are starting to mature a little bit as well. Facebook is moving through that pretty quickly and Amazon is trying to do a whole bunch of new things with their Echo platform and the Alexa personality.

So what we see now is that most chatbots have a very structured and regimented way of communicating. It looks for keywords, commands and trigger points, whereas the natural language processing of Gwyn really allows you to have a conversation. As we adapt, and more and more companies start to figure out what is the best way to engage customers, everybody benefits from that. But the front-runners benefit from it the most.

Subscribe today for just $116 a year

Get the very latest news and insight from Campaign with unrestricted access to campaignlive.com , plus get exclusive discounts to Campaign events

Become a subscriber


Don’t miss your daily fix of breaking news, latest work, advice and commentary.

register free