Virtual assistants and chatbots are here to make our lives easier. Want help finding gifts? Retailers ranging from Nordstrom to DSW and Sephora offer chatbots to find presents for your friends and family through Facebook Messenger. Need advice on making a dessert that’s gluten free? Whole Foods’ bot makes it easy to find a recipe for any occasion and dietary preference. Or if your hands are full, Alexa can do the same thing through voice commands. Whether you’re shopping, cooking, or just looking for entertainment, there’s a bot for that.
At their core, however, bots are simply applications that perform a designated task. And like any application, it can be good or bad at its function. Marketers strive to create chatbots that add value and contribute to an effortless experience that surpasses a mere novelty or fad — but chatbots are still in the experimental stage.
“Chatbots offer immense potential for consumers to interact with a brand in an organic way, through one-on-one conversations within the messaging or social apps they already use,” says Brian Seewald, vice president of digital at DSW. “There are many nuances to consider when building a chatbot. AI-driven technologies are the first big step in making successful chatbots. That said, these bots will also require thoughtful planning to ensure bots are designed to easily fit in consumers’ lives and enable them to interact at their fingertips, whenever needed.”
Beyond the fundamental layers of data and software, marketers need to plan how chatbots will fit into their broader customer engagement strategy. When creating a bot, there are some key principles to create a memorable, easy-to-use bot that will help ensure success.
1. It’s created with (one) purpose
After the mobile app revolution of the 2000s, we quickly saw an ecosystem evolve where there’s an app for everything. Similarly, now we’re saying, “There’s a bot for that.”
In these early stages, however, it’s essential that bots don’t aim to be a solution for everything. They should set clear expectations with users about what they can do — and more importantly, what they can’t do. Otherwise, brands will find themselves facing escalated customer frustration as users ask for things the bot isn’t set up to handle. A positive example is 1-800-Flowers’ Messenger bot, which clearly states that the bot helps users order flowers and other products. The mission is direct, clearly communicated, and created for a particular customer need. Over time, the company plans to offer features to notify users of special occasions, send Facebook messages letting the recipient know a package is on the way, and share an update when the flowers have been delivered.
In contrast, look at CNN, which launched a bot it claimed would answer questions about the news. Because of the breadth of possible questions, however, the bot often failed to understand what users were asking and didn’t provide the information that users expected. Google Allo addresses this issue with its chatbot by letting users know if a query is too complex to answer and suggesting alternate questions to ask, continuing to provide value. Setting clear expectations and giving users advice on how to interact with the bot — like specific phrases or guided prompts — will lead to a win-win for the user and brand.
2. It knows you — within reason
When you’re talking with a sales associate, you expect them to understand your needs without asking you to repeat yourself. Shoppers expect the same of chatbots. Like a good sales associate, a bot should remember your previous conversations, purchases, and preferences. This builds contextual understanding.
By using progressive profiling, bots can gradually learn other relevant information over time about what you want — and what you couldn’t care less about. A smart experience creates a big picture of your behavior over time instead of going for a big bang approach, so the overall quality of interactions is smooth.
A bad bot asks the user to fill out a lengthy profile up front, creating a significant barrier to adoption. This is an onerous task for the user, and it’s hard for most new users to justify investing their time and trust without getting a taste of the potential benefits.
There’s a fine line between contextual profiling that’s smart and that’s just creepy, and marketers should take care not to push the boundaries. A good bot doesn’t ask for personal information (like contacts, address book, access to photos or files, etc.) unless it clearly explains why that information is important to the experience. A bad bot attempts to exploit the user and harvest their information to invite others in a “growth hack” way. This will destroy any semblance of trust, not just with the bot but with the brand.
3. Speak the same language
Talking to a bot should be as close as possible to the organic experience of talking to a live person. An engaging bot balances binary questions with effective natural language programming to make the conversation flow at a natural pace. In contrast, an ineffective bot makes the user answer a dozen yes-or-no questions, leading to “survey fatigue.” It’s OK to insert call-to-action buttons and user-interface styling sparingly, but reserve such touches for the most significant decisions (votes, actions, etc.). By only asking binary questions, poorly designed bots make the entire experience feel transactional and crude, essentially wasting the capabilities of the bot and leaving the user with a less sophisticated image of the brand behind it.
Most users who interact with a brand’s bot love the brand experience on other channels and expect the bot’s tone to be consistent. The bot acts as a brand ambassador, so don’t compromise on the personality and voice.
While chatbots are used primarily as digital assistants and customer service agents today, we’re only in the early stage of innovation. AI will continue to advance and inspire experimentation among marketers. It may not be long until we have chatbots acting as teaching assistants, health care support, or even financial advisors. Wherever the future may take us, the most effective chatbots will be the ones that merge technology innovation with existing customer engagement strategies.