chatbot_lawyer_success

Update, May 4th 2018: The guys at DoNotPay have changed things around! Though the screenshots below are out of date, the lessons are more relevant than ever.


The so-called 'chatbot lawyer' made a lot of noise in the press.

Joshua Browder built the chatbot lawyer in attempt to help 'the most vulnerable people in society'. Using artificial intelligence, the chatbot helps users contest parking tickets through a chat interface.

Almost a year ago, Joshua's chatbot had reportedly taken on over 250,000 parking ticket cases, successfully winning 160,000 of them.

The story was widespread in the press and media. For many, this was their first encounter with a useful chatbot. A 64% success rate (and over $4million in parking tickets appealed) is, without a doubt, impressive.

But, what made this chatbot so successful? What can we all learn from this story to make us plan our chatbot builds better? This is exactly what we are going to try to figure out in this article!

 

What does the chatbot lawyer do?

First things first, let's look at the actual usage of the chatbot.

I impersonated a driver who just received a parking ticket and wants to appeal. I head over to DoNotPay.com.

chatbot_lawyer_welcome_screen.jpg

The world's first chatbot lawyer makes it super easy to get started. Just enter your email address and a password, and you are good to go.

(As a one time user, I fail to see the appeal of an actual account creation. Also, why email? But let's skip that for now)

I create my account and am taken to a new page. On here, I see the options I have. I pick DoNotPay Robot Lawyer.

chatbot_lawyer_chosing_screen.jpg

Finally, I am presented with the chat interface and greeted with a welcome message. Although the interface is slightly unorthodox, I quickly understand where I should start typing.

chatbot_lawyer_welcome_message.jpg

I follow the bot's questions and answer using words. I am quickly prompted with a link to generate my appeal - success!

chatbot_lawyer_conversational_ux.jpg

(I purposefully used a simple scenario and did not try to break it. The purpose of this article is to see why the chatbot lawyer is successful, not how to break it!).

I am asked to fill in a few more details about my infraction to generate my appeal.

chatbot_lawyer_more_info.jpgFYI - my details are made up.

The appeal is generated for me using the details I have entered, and supported by relevant references to similar previous cases.

chatbot_lawyer_appeal.jpg

I can now print and send it to the right company.

 

What made the chatbot lawyer successful?

Now we're all up to speed on how the bot works from a user point of view, we can focus on the purpose of this article: why did the chatbot lawyer become so successful?

I believe it is due to five factors.

 

Simplicity

The chatbot is extremely simple.

It is simple to use. It is simple to understand. It is simple to read. From a user point of view, there are close to zero barriers to entry.

The only potential tripwire is the initial request for an email & password.

Once you are past this point and land on the chat interface, you are smooth sailing until completion.

Lesson: simplicity wins. Using a chatbot should not be complicated. Make every part of the process as clear as possible, and reduce the number of hoops your users have to jump through before experiencing the real value of your chatbot.

 

Great conversational UX

This aspect partly falls under the previous point about simplicity.

Conversational UX is an enormous part of the success of any chatbot. It is the way your chatbot engages its users, how the conversation flows, the words it uses, the sentence construction, and more.

Read more about conversational UX.

Here, we can see how the conversation flows extremely well. I didn't need to think too much about the sentences that were thrown at me. I was even able to give one word answers to each question without hardly a thought.

Think about it. This is a lawyer chatbot. Traditionally, law does not have the easiest to understand lingo. This chatbot could have been a whole lot worse had the engineer behind it tried to copy a 'legalease' line of questioning.

Lesson: you may know a lot about your product, but your users may not. Make sure the conversational UX of your chatbot suits the people who are going to use it. Keep the lingo to a minimum.

 

The one true goal

Easy one right here. The chatbot lawyer has one goal and one goal only; to get you out of that nasty parking ticket.

It never strays aways from that goal. It doesn't try to send you pictures of silly road signs to make you giggle. It doesn't do 'small talk'.

crazy_roadsigns.jpgA picture of crazy roadsigns? Not necessary. Stick to the goal.

It is focused on its one goal.

This not only makes the chatbot extremely easy to understand for its users, it is also what makes it super-successful. No one gets lost during the conversation. No one gives up. No one uses the bot 'just for fun'.

Lesson: the one true goal rule is real, follow it. What is your chatbot's one true goal? Is it to sell more shoes? Make it sell shoes. To answer questions about your products? Make it do that.

Focus on achieving your goal first, everything else is fluff.

 

There is a real need

The biggest secret behind the chatbot lawyer's success? It is solving a real problem.

I am sure plenty of people every day get parking tickets and feel they have been wrongfully accused.

Yet, most people do not appeal simply because it is a hassle and might require spending money. Paying the ticket ends up being less hassle, even if it is unfair or wrong.

The AI-driven chatbot is the answer to this very hassle. It makes it easy for anyone to appeal. It literally takes a couple minutes, does not require any legal knowledge, and is entirely free.

Who wouldn't want to try it?!

Lesson: to make something successful, most of the time you have to make it useful. Save your users time, money, hassle, or something. Make your chatbot the best way to achieve something.

 

An inclination for virality

Finally, there is a propensity for virality in this chatbot.

Not so much during the actual process of using the chatbot, but when the appeal has been successful.

Imagine using the chatbot to successfully avoid paying that £160 parking ticket you feel was wrongfully sent to you. Of course you will wax lyrical about it.

This is most certainly what drove a lot of the adoption for this chatbot. Every successful appeal turned into more shares, more virality, and more use - thus more success, and so on.

Lesson: build in virality in your chatbot. But, don't just focus on gimmicky 'share' buttons here and there. If you have followed the previous pieces of advice, a successful interaction with your chatbot should be all the viral nudging you need!