It’s not just robots that are out to get our jobs. The chatbots are coming too. In fact they’re already here, taking over your company’s email, advertising, and every other kind of marketing!

No, not really.

The chatter in some circles is that bots are killing off traditional channels, but it’s not quite true—at least not yet. To understand how bots actually fit into marketing plans and communication channels, let’s take a closer look at what they are and what they can do.

The basics of bots

When most marketers think about bots, the first type that comes to mind is a chatbot, which have been successfully launched across a wide number of platforms like Facebook Messenger, WeChat, Slack and Kik, as well as within apps and sites. Chatbots have grown quite a bit in popularity over the past year as brands and marketers experiment with them. They can be something as simple as a conversation script that defines how a bot should engage with, and respond to, a consumer (think: find a free beer coupon for local bars). Or they can be much more complex, leveraging artificial intelligence to replace the interaction with a customer service rep.

Text-based interfaces aren’t the only type of bot used by brands, however; they can also be created for voice-enabled devices like Amazon’s Alexa family and Google Home. The structure of chatbots and voice-enabled bots is fairly similar, I learned recently from Malcolm Paul at app development agency BKLYN, who pointed out that what you’re programming first is the interaction between the consumer and the brand. So, if you’re considering developing bots for multiple devices and channels, it’s likely more cost effective to create them at the same time since the basic structure for each starts in the same place.

Of course, there are different considerations for building a voice-enabled app or bot. While text-based bots allow you to deliver links and images within the conversation, voice-enabled bots must be much more precise in their answers since there are no visual cues. And voice tends to work better at home, where you’re more likely to be multi-tasking yet feel comfortable speaking aloud to a device—something that’s still a bit awkward to do in public or at work.

What are bots good for?

At a recent Digital DUMBO panel on bots, Havas Executive Digital Director Sami Viitamäki described chatbots as nicely complementing the existing marketing landscape, given their immediacy and personal nature. Beyond the traditional definitions of know, go, do, and buy, however, he suggested that “connect, grow and play” could be powerful consumer motivations for interacting with bots in social spaces like Facebook Messenger.

One recent example of integrating a chatbot into a mobile app is the Warby Parker Prescription Check app. At the beginning of the app experience, it contains a chatbot that helps consumers do the refraction test, which is something that normally happens in the doctor’s office. “The thought was that it has applications for rural areas, so by making it really approachable it could be brought just about anywhere,” said Lauralynn Drury, Strategy Program Manager at Warby Parker. The purpose of the bot is to ask a series of questions to evaluate the risk level of using the app; if a customer has certain risk factors, the bot will recommend they be seen by an eye doctor instead.

The increasing brand interest in voice-enabled bots and apps is being driven by two chief factors: consumers’ growing comfort with voice-driven artificial intelligence, as well as an increase in voice-enabled device sales. According to a recent Business Insider article compiled with eMarketer data, an estimated 35 million Americans use a voice-enabled speaker; Amazon’s Echo dominates with a 70% market share with its family of devices that run Alexa. Amazon provides a Skills Kit for developers and brands to create their own apps on the platform—there are now more than 15,000 available—and Google followed suit by making its Actions available to third-party developers.

More importantly, Google made its Assistant software available as iOS and Android apps in May, greatly increasing the reach of Actions that are developed to run within the Assistant. Why does that matter to brands and developers? While Skills apps are limited to the owners of Alexa-powered devices like the Echo, Google’s Actions are housed within an Assistant that aims to take on Siri, the iPhone’s built-in virtual assistant. That could mean millions more users for the voice-enabled apps you develop.

Defining the ability of bots

Bots are great—when they work. But there are lots of examples of bots that don’t work well, misunderstand questions, or just revert back to the same short script when stumped by a user’s input, causing frustration and negatively impacting your brand.

There are essentially two routes you can go, and It’s a debate within the industry which will win out in the long term: Limit the bot to few relatively safe options without a lot of room for interpretation, or utilize artificial intelligence to try to replicate the experience of speaking with a human. The latter uses natural language processing, known in the industry as NLP, which is especially important in fielding unexpected questions or errors.

Do you need natural language processing? Not necessarily, but if you do use it you need to invest in it fully. For a simple bot that has a narrowly defined purpose, many developers eschew NLP in favor of a simpler build; in those cases the best thing to do is frame the bot up front by having it state what it can and cannot do, setting consumer expectations for how to interact with it properly. Equally important, however, is giving users an escape route, even if you’re using NLP. Especially for customer service related bots and apps, there should be a way to contact a live human for help—ideally one who can see the history of the chatbot and pick up where technology left off.

Considerations for getting started

First, don’t build a bot just to have a bot. Even funny bots that eat emojis or deliver cat videos still serve a purpose. You’ll need to sit down with your team and figure out what customer problem is really solved by a bot: are people getting stuck in an onboarding process, or dropping out of the sales funnel? Is there a service or experience that could be improved with a bot? How could your brand’s bot, app, Skill or Action add value to a customer’s life?

Next, there are options in terms of platforms, which should help you narrow down the type of developer partner you’ll need. Some companies build bots within existing apps so they can get user behavior, demographics and other rich information to improve the experience. Others rely on existing messaging platforms such as Facebook Messenger, which allows you to tap into that 1.2 billion user base and leverages the bot discovery tools launched last week. And still others specialize in voice-enabled technology, or more than one platform. It also depends on where you’re expecting to go in the next 2-3 years; depending on your goals, you may want to invest in a more robust infrastructure like IBM’s Watson.

Finally, as you’re evaluating developers and partners, your internal team should think about how you’ll measure success, which is closely related to the bot’s purpose. Measurement may include such metrics as:

Followers, downloads and impressions
How far people get through the process
How many people complete a certain action
How many times people interact with the bot
How many people engage when the bot sends a push notification

Depending on what you’re aiming for your analytics will differ, but knowing what your goals are and keeping those as your north star will help you stay on track throughout the development, user testing and launch phases.

Have a question about branding or another marketing topic? We'd love to help: just ask!