Customer service chatbots miss the big picture
Most of us are building bots the same way we built apps and sites: We are still building products for groups of people. But bots finally have a superpower. They can now personalize and customize content in a scalable way. Websites and apps are meant to solve problems for groups, but bots can solve problems for individuals.
Websites essentially cram entire product lines on the homepage in hopes that it will appeal to you. A great example of this is the Polo Ralph Lauren website. Even when this is done to perfection, a great site will have a 40-60 percent bounce rate and a 2-5 percent conversion rate. This means that about half of your traffic is lost immediately and 95–99 percent does not convert. This is a huge waste.
Most products have a goal of solving a painful problem for a specific group of people. Great companies learn a lot about their customers and build entire solutions for them. This is done through questionnaires, interviews, and A/B testing. The result is an average of what most customers might do, and the final result becomes a generic flow, a one-size-fits-all type of solution. Many shopping bots also try to cram in a full product line, 10 words at a time. In the end, it feels like a really bad game of 20 questions.This is what happens when you try to condense a website into a bot and rely on averages.
In the end, if you don’t know why a person is on your site or using your bot, you end up playing a guessing game and hoping that the shoe fits. Most of the time it does not. The worst part is that most bots cram in information every time you use them. In order to get to the useful parts, you need to go through a maze. This way of organizing information is a broken activity. It barely works online, and it will surely fail in bots. Fortunately, there is a much better way.
Bots offer us a better way. Instead of A/B testing a product with groups of people and arriving at an average solution, bots allow us to do A/B testing with individuals and offer the best personalized solution possible.
“Great bots won’t have one or two flows, they will have multiple flows and copy each tailored specifically to a person’s preferences,” says Aparna Sharma of Chatbot’s Life.
Bots are great at getting information from people. In fact, it is one of their specialties. When it comes to getting to know a user on an individual basis, a bot has the advantage hands down. The best way to take advantage of this is by asking meaningful questions and creating a context for your user. For example, if we built a bot for Ralph Lauren, the goal of the bot would be to become a personal shopping assistant for the user. This means that the bot has to learn the customer’s preferences and habits to help them shop better.
Currently, there is an easy, no-programming-required way to do this. Chatfuel does this on the front end via segmentation, and Botanalytics does this on the backend via re-engagement. In both cases, you will be leveraging user inputs and data to push relevant personalized content to your users.
“For the first time, technology is offering a scalable way of having personalized 1×1 experiences with your users,” says Dmitry Dumik, CEO of Chatfuel.
The power of the right questions
The Christina Milian chatbot leverages the power of questions to personalize flow, content, and notification. One of the questions the bot asks users is if they are “taken” or they are “single.” The bot will then take this relationship status into account, and the entire flow for each user group will be completely different. In the end, a bot should not have one user flow and copy; it should have many flows. The very act of engaging with the user is an A/B test.
The second way to do this is via Botanalytics. By looking at how different user groups are engaging with your bot, the types of content that really engage them, and where they drop off, you can quickly see what works best. Based on this, you can create rules, put users in different buckets, and re-engage with them based on their individual interests. By using this technique, PennyCat increased engagement by 87 percent!
“Soon, we will be able to accurately predict a user’s behavior and automatically suggest the message they are most likely to act on and the best time to deliver it,” says İlker Köksal, CEO of Botanalytics.
This means that every person will have a different experience with our Polo Bot. The bot will know our shopping habits, which types of Polos we like to buy, the best price points, and our location. It will then offer the exact right deal when we are most likely to buy. This is the new paradigm. This is when bots have superpowers.
This story originally appeared on chatbotslife.com.