What Nike's chatbot taught us about conversation design best practices

We recently kicked off Breaking Bots, our new workshop series where conversation designers deconstruct a well-know chatbot, then put it back together better—with lots of commentary and audience participation—live. Two wildly talented conversation designers joined us for this first one: Ali Pinch, Conversation Designer at one of Australia’s largest retailers, and Nicolle Merrill, Senior Conversational AI Designer at Intuit.

If you weren't able to join us, we've highlighted the best practices Ali and Nicolle shared over that hour (and for even more takeaways, read this recap post). To see the breakdown and buildup unfold in real-time, watch the episode below.

1. Start with your unique business context

Every business has unique needs, goals, and audiences. This context is critical when designing chatbots. Consider how a retail business and healthcare provider might differ. The retail company may want to help customers track their orders, whereas the healthcare provider may want to help patients book appointments. The company’s goal in each case would impact how these bots are built and trained. That’s why it’s critical to start with an in-depth understanding of your unique business context.

2. Keep customer intent top of mind

Understanding the intent of your customers helps you empower them to get what they need as fast as possible. Nike’s team could have benefited from this mindset. When Nike’s bot asked customers if they wanted to see new or existing products, it created confusion instead of clarity. What the customer wanted was probably to figure out if a product was in stock at a store near their location. Instead, they were left scratching their heads, which delayed them from getting the information they needed.

3. Get people what they need as fast as possible

As we just implied, your goal should be to get people what they need as fast as possible. In the Nike example, one journey took five steps instead of two. This fosters frustration instead of delight. As conversation designers, it’s our job to remove friction from the user journey. Remember: The best designs are simple.

4. Ask for only what you must have

Chatbots are often built to reduce the cost of customer support. While it can be tempting to collect contact information for marketing purposes, that creates more work for the customer, which can cause frustration and even reduce loyalty. That’s why we recommend asking for only what you need to serve the customer. In Nike’s case, the bot asked for the customer’s name and email, when it only needed an order number to process their request. By asking for multiple forms of identification, it wasted the customer’s time and tested their patience.

5. Communicate the bot’s understanding

When customers interact with a machine, they may have doubts about whether that machine understands what they mean. For this reason, it’s important for your bot to repeat back what the customer has said. Say a customer asked Nike’s bot if Dunks were available at their Melbourne location. The bot should respond by confirming that it can look up whether Nike Dunks are in stock at a location in Melbourne and, if necessary, ask the customer to select a store in that location.

6. Solve for 80% of your audience

Unless you aim to create a complex chatbot with a large number of potential pathways, a good heuristic to keep in mind is to solve problems for 80% of your audience. For example, retailers like Nike may want their chatbot to ignore certain variables, such as price and size, which have infinite possibilities. Instead, they may want to direct customers to the relevant product page to view that information.

The importance of a delightful user experience

The key takeaway from our first episode of Breaking Bots is the importance of fostering a delightful user experience. At minimum, your bot should use clear and concise language, cater to customer intent, and help customers get what they need, fast. You can improve the user experience further by leveraging an engaging brand voice and tone and offering encouraging feedback and positive reinforcement. Together, these tactics can help you boost customer satisfaction and loyalty while driving down your operating costs.

To see the breakdown and buildup of Nike’s bot unfold in real-time, watch the episode on YouTube.

Header photo by Terrance Barksdale.


Breaking and rebuilding Nike's chatbot

No items found.