What Chatbots Can (and Cannot) Do for Retailers 

Exploring capabilities and limitations

Author: Tricia Carr

July 21, 2017

Chatbots have the potential to influence consumers at many touchpoints on the path to purchase. Bots are often used by retailers to feed personalized product recommendations to a customer via messaging platforms like Facebook Messenger. But do these bot interactions actually generate purchases? Jeff Malmad, managing director and head of Life+ at Mindshare North America, spoke with eMarketer’s Tricia Carr about what chatbots can and cannot do for retailers.

eMarketer: Many retailers are approaching chatbots as a new channel for providing customer service or offering personalized product recommendations, but do bots have long-term potential in these areas?

Jeff Malmad: We're really bullish on bots. Messaging with the masses in a personalized opt-in environment is critical, especially from a customer service perspective.

Customer service is a necessity, and consumers want to be at the core. They will go to any lengths to get at the core and sometimes express their dissatisfaction publicly. A bot is a great way to begin that dialogue in a nonpublic environment.

eMarketer: Do shoppers like interacting with retailers via chatbots? Is it in any way similar to chatting with a friend?

Malmad: Messaging has a psychological effect. In a conversational interface—say you and I are chatting over iMessage—you can see those little dots [that let you know the other person is typing], and that triggers your dopamine receptors. Consumers are programmed to seek information and be rewarded immediately. That’s why people are always texting—they expect an immediate reply. That's why people get excited when they get a “like” on Facebook or Instagram. This mentality plays a large role within the world of messaging.

“Consumers are programmed to seek information and be rewarded immediately.”

eMarketer: Is engagement typically high when consumers interact with retailers via a chatbot?

Malmad: There is higher engagement and a longer time spent when you provide consumers with a conversational interface that allows them to have a dialogue. But it has to be done in a way that allows consumers to opt in to the conversation and allows you to provide them with incremental value as they go back and forth with the bot.

eMarketer: Do you have an example?

Malmad: We worked with a retail client during the holidays, and we wanted to help consumers navigate through their holiday shopping needs with a customized, pre-populated experience within the environment on Facebook and [messaging app] Kik.

The overall program performed excellent. The average engagement per user was eight messages going back and forth with the retailer. They spent over 25 seconds within the bot experience, and after engaging with the bot and receiving relevant holiday product recommendations, there was a 75% chance that they then navigated to the retailer’s site.

eMarketer: What will it take for consumers to feel comfortable making a purchase during or immediately after a bot interaction?

Malmad: Right now it's all about the curiosity factor and understanding whether or not this bot can provide the customer with something they truly need—or if it’s just a way to get the customer to the retailer’s homepage. We approach chatbots as curated experiences that provide users with the best options based on a series of questions that we ask you. It’s like a personality test. It’s a fun, interactive experience.

“Many of these experiences are based on ‘if you do this, I do that.’ It's not true artificial intelligence.”

Many of these experiences are based on ‘if you do this, I do that.’ It's not true artificial intelligence. Right now we're leveraging automated interfaces, but once there is artificial intelligence and true dialogue with a retailer in a bot experience, it’ll become much deeper and much more personalized.

eMarketer: If the user doesn’t make a purchase right after a bot interaction, how can the retailer keep them moving down the purchase funnel?

Malmad: We want platforms and bot companies to allow us to take the information we gathered and use it to re-message the customer somewhere else. If a customer had a dialogue with a retailer about a specific clothing brand, you could re-message to them when they’re on a different site later that day to either get them back into the funnel or get them back in the mindset of purchasing their products. Or if the customer went as far as putting something into their cart, how do you get them to convert that day? There are a lot of tech companies focusing on this challenge.