A “C-L-U-E” To Improve Your Bot Metrics
Despite all the hype and promotion from bot enthusiasts, many bots (or automated response agents) have struggled to find an engaged and returning audience. After measuring millions of conversations across several hundred bots on Facebook, SMS, Kik and Slack on Botmetrics, we’ve gained key insights on what separates the best bots from the also rans.
We packaged these insights into a C-L-U-E framework that bot-makers should consider to build bots that customers love.
TL/DR; Start with a clear Use-case or “a job to be done.”
Effectively gather enough user Context to do the job, complete the job with Love and empathy for the user and finally set clear Expectations to help automated agents deliver outstanding service to customers and become a part of their workflow.
Let’s dive in.
Context
The C in C-L-U-E stands for context. Like your friend who asks you what you do each and every time you meet them — users quickly tire of bots who don’t remember basic information about them.
The majority of the interactions that your bot supports typically need to have or maintain user context.
This could be as simple as remembering that the user is still looking for a hotel in Atlanta when they ask to change the dates in a hotel search bot.
The Domino’s bot takes this to the next level by pulling the customer’s order history and allowing them to 1-click re-order their most recent meal or their favorite meal.
Understanding and accessing the user’s context will make your conversations smoother and memorable.
Love
Second, build your bot with Love — for the user. In UX speak we call this empathy. There are three areas where we see bots fall short on empathy:
Never ending questions: If you’re asking for an address then simply ask for the address. There is no need to create a laundry list asking for street address, city, state, and zip code separately. Intelligently chunking data allows you to get the information that you need without an endless barrage of questions.
Notifications: First allow users to opt-in to specific notifications. Then use them rarely if at all. In messaging, your bot’s messages get a trusted place beside friends’, coworkers’ and family’s messages. Abusing that trust quickly leads to bots being disabled and blocked.
Slack, a trusted leader in messaging, uses a complex flow chart to decide wether to notify a user or not. If your Bot is not being as skillful in using notifications, you should consider whether you should uses them.
Charming Is Better Than Cute: Most bot makers spend a lot of time giving their bot a lot of cute personality flourishes.
While this is cute the first few times, like your old uncle Tex’s eccentricities it gets tiresome really quickly — especially in failure cases where the user is frustrated.
This is made worse when bots use humor to deflect from a failure in UX, NLP or functionality. A better approach is to focus on adding personality by being charming rather than simply cute. In practical terms this means completing the task at hand and then going beyond the job at hand in a surprising or unexpected way.
Use Case
Define a clear Use Case for your Bot — What problem are you actually solving? This means truly understanding the problem you’re solving and why conversational UI is a better than the current solution. Simplify not simply re-invent.
Expectations
And finally, set clear expectations. One way to do this, as Des Traynor of Intercom says, is to focus on building a scalpel rather than a Swiss Army knife.
Set clear expectations on what you bot can do and then over deliver on delighting your customers within those expectations.
You don’t walk to you local deli counter and expect them to indulge you in a tap dancing routine. Similarly, most users don’t expect your bot to tell jokes and fetch the weather.
When DoNotPay the Robot Lawyer started, it began by doing one thing: helping people contest their parking tickets in the UK. After proving it’s value, the bot earned the right to offer other legal services to it’s customers.
The C-L-U-E framework provides a robust way to think about the “job that you bot does.” Understand why your users are using your bot and make that experience delightful. If you do, you will automatically create engaged users that continue to “hire” your bot.
