In this post, I will briefly talk about RoboRobo’s latest chatbot. Our riddle game bot, which we named Agymozgató (Mind-Breaker) has achieved very good results in just a few days, and we have also gained a lot of exciting experiences, so I think it is worth mentioning here too.
The goal of the project was primarily to research Bot UX and to study user behaviour. After the more function-based Sihuhu, this time we wanted to make a lighter, playful but to entertain users (and ourselves too 🙂 ).
How does it work?
Using the bot is really simple, although its cunning puzzles can challenge the players.
In any case, we wanted this tiny brainstorming to end up with positive feelings and success for our users, so we have considered the importance of including the following two things: the ability to ask for help and a rewarding system. I’ll have a few more words about these later.
🤖 Bot personality
The target audience was primarily the 18–25 age group (they are the main user base indeed, although users from older age groups also make a significant number), so the communication of the but is dynamic and its wording is youthful.
It is important to emphasize that the bot does not do chit-chat with the players, its operation is clearly solution-oriented. As a result, users will not be able to divert messaging in any other direction, and in any case, they will visit the game for nothing but the enjoyment of the brainstorming.
How the riddles work was something that could be transferred all in one to the chatbot, so the game is almost exactly as in real life.
Given two people, one of them asks the question (in this case, it is the bot), and the other makes guesses and tries to figure out the answer (this is the user). If the player is lost, they ask for help from the other person, who tries to bring the player closer to the solution in a few words (just like what the bot does).
Agymozgató is also as flexible as a real person in the matter of accepting the right answer. It is not important for the bot that the users write their answers grammatically correct or totally accurate, as the bot ”thought”. The point is that the chatbot can see that the player has recognized what the answer may be, so the system accepts close synonyms or different spellings of the correct answer.
🔄 There’s no stop!
For real! The user always has an active question, and as they guess or skip it, the bot will immediately send another question. Riddles will be sent in random order so sometimes the riddle is a bit more difficult, while another time it gives a very easy question. It is refreshing to solve a simple puzzle after a more complicated one.
ℹ️ We help if needed
Users can skip questions at any time. To do this, they need to either send a message (skip, another one, I give up, etc.) or click the “Skip” button from the persistent menu.
There are 2 pieces of help for each question. If the user wants to give up, the game will offer to send help in exchange for a star. Users are happy with a tiny piece of information to help solving the riddle. Because it costs a star, many people will consider stars valuable, and if possible, they’re going to try to solve the challenges without help, to save their starts they worked for.
The game offers help after the 3rd, 6th and 12th incorrect answers. We wanted to avoid frustration when users meet with a bit more difficult question.
If the user has already asked for the two lifelines per question, the bot will remind them that they can give up anytime.
Rewarding and motivation
While designing the UserFlow, we consciously sought to motivate and reward users. There are 3 types of rewards in the game: GIFs, stars and praises.
After the user gives a correct answer, we first send a humorous, positive vibe GIF. (Fun fact: we’ve gathered over 50 different gifs to make sure users don’t get bored). We have sorted the gifs so that all ages can understand the message of them; we avoided overusing memes.
Users can get stars for the right answers. These stars can be used to measure their performance. The number of stars given depends on the difficulty of the riddle and, in addition to being collectable, they can be spent for help; each costs 1 star.
🏆 Praising message
Finally, after reaching a certain amount of scores, we end up with a personal rewarding praising message. Based on feedback, the latter especially melts our users’ hearts!
During the project, besides the development of the chatbot, we also paid special attention to communication. Not only we posted on Agymozgató’s Facebook page, but we strived to establish a continuous and direct contact with our users. In all cases, we responded to questions and problems that arose, taking care of the friendly and youthful tone.
In some of our posts, we encouraged users to interact, which is a huge reason for the rapid organic spread of the chatbot.
We are very pleased with the results of Agymozgató, and we can say that we have met our expectations too. At the time of writing, the bot has been running for 14 days, and statistics shown below summarize the results of the past 2 weeks.
#️⃣ In numbers
The number of messages received has now exceeded 5 million, and the number of outgoing messages is over 10 million.
The Agymozgató Facebook page currently has more than 4,000 followers and, with over 900 reviews, has a 5-star rating. The presence of users is continuous, with more than 32,000 players in total, and 5–8,000 active users per day trying to solve riddles.
At one point in the game, users are asked to rate Agymozgató, if they have a few minutes. We have been rewarded with lots of 5-star ratings. Here are some of our players’ reviews:
Agymozgató has proven that the forms and ways of chatbots are used is only limited by our own imagination and that users are happy to play with text-based games.
Check it out yourself: http://m.me/agymozgato/
Disclaimer: this is a translated version of an article originally posted on our own botblog.hu, on February 7, 2018.
Bot stats as of March 22, 2019:
32,000,000+ incoming messages, 190,000+ unique users, 5.0/5 rating with 1,500+ reviews.