Chatbot Users Have Higher Purchase Intent (If You Engage Them the Right Way)

An interview with Betabrand's Aaron Magness

Author: Tricia Carr

June 22, 2017

Rather than build its own branded chatbot and lure consumers to interact with it, crowdsourced apparel company Betabrand partnered with fashion chatbot Epytom to engage with its existing user base. 

eMarketer’s Tricia Carr spoke with Aaron Magness, Betabrand’s CMO, about how many users are willing to purchase something recommended by a chatbot, and what brands should be wary of as they test the technology.

eMarketer: What was the biggest challenge for breaking into the chatbot space?

Betabrand's Aaron Magness

Aaron Magness: Chatbots allow you to engage with the end-user in a way that’s quite different from how they normally expect to engage with a brand, but how they’re comfortable engaging with other people. How do we marry this together and have a conversational approach that actually provides value?

A lot of brands quickly built bots that try to replicate their websites, but in a bot version. We took a slower, test-and-learn approach to figure out what resonates with the end-user. We aren’t just replicating the standard shopping experience.

eMarketer: Betabrand partnered with the Epytom chatbot, which delivers outfit suggestions to its users on a daily basis via Facebook Messenger. What was the strategy behind this move?

Magness: We saw an opportunity to do something different with Epytom vs. doing what everyone else is doing. Epytom helps busy people [decide what to wear each day], and our core customers tend to be busy professionals. [Epytom] already has an audience of engaged users. We started working with them to engage with people with that type of personality, and spread the message that Betabrand has clothes that fit their lifestyle.

eMarketer: What have you learned so far from this partnership? Are Epytom users willing to purchase items they’ve discovered from a chatbot?

Magness: It’s early and we’re still testing, but there are a couple of learnings. The people who engage with the Epytom bot and then come to Betabrand to shop are almost exclusively new users. It’s driving a new user base to our site. The users who click through to shop have a higher conversation rate than our standard mobile conversion rate. If you engage people and provide them with the right product [in the right environment], there is a higher level of intent.

eMarketer: What does this say about these customers? Do you think the conversation rate will start to level out?

Magness: To put it in perspective, it’s a relatively small customer base. People who engage [with chatbots] tend to be early adopters. The data we’re pulling on usage and engagement will help us as we go forward. We’re interested in knowing how the people who have made a purchase behave after the fact.

“Just because [bot users have] a very high conversion rate now does not mean that everyone who uses the bot will have that same conversion rate.”

We’re also conscious that there can be false positives. If you isolate your best customers on mobile in one environment, they’ll have a higher conversion rate than the average. You don’t want to say that people who gravitate toward engaging with bots are going to be your best users. They are probably going to look like your best users, but just because this cohort has a very high conversion rate now does not mean that everyone who uses the bot will have that same conversion rate.

eMarketer: What other challenges are you facing now that you’ve found an audience in the chatbot space that’s willing to purchase?

Magness: When you’re shopping in the apparel space, you start with a category like bottoms, tops, outerwear, dresses, etc. Then you get down to color and size—that becomes a lot of information. Whereas in beauty, for example, it’s a binary choice: Do you want this red lipstick or not? The decision happens a lot faster.

When you have to gather a lot of information in the bot environment, the communication starts to break down. But when you can make the conversation that leads to a purchase very simple, it becomes more effective. If someone bought a pair of our Dress Pant Yoga Pants, we already have their size, shipping address and payment information. We can ask, “Did you know we have them in additional colors? Do you want to order more?” and the user can respond “yes” or “no.”