Multiplatform chatbot for users of the PMI tobacco heating brand

Client

Brand tobacco heating system from Philip Morris International

Context

Until 2020, the main digital channels for involving customers in brand communication were the website, informational email and SMS mailings. According to Philip Morris International, the brand’s products are used by more than 3 million adult customers.
Since most of the brand’s audience is used to receiving information in a mobile format, the company decided to introduce a chatbot to the main digital communication platforms.

Task

The online assistant had to solve the following tasks:

Improve the user experience;
Reduce the load on the support service;
To ensure constant communication of users with the brand – 24/7 and in any user-friendly interface.

Solution

1 To test the demand chatbot as a channel for communication between customers and the brand, we have launched a pilot online assistant in Telegram.

In the “zero” version, the chatbot represented an MVP with two buttons – “start a session with an operator” and “end the session”. The bot joined the operator channel in Webim, the operator connected to the dialog, which communicated with the user.

The online assistant attracted the attention of users, so the company decided to develop and increase functionality in order to retain the audience and make the service as useful as possible.

2 In the next release, all the basic features of the bot became available to users: ordering a tobacco heating device for a test drive, a library of materials on how to use and independently diagnose the device, requesting an individual referral code and the ability to contact a support operator.

In parallel with the development, we have set up data collection and end–to-end integration with the internal services of the brand – since the bot is located in the highly loaded circuit of affiliated companies of Philip Morris International.

3For users, the bot’s interface looks as simple as possible, but “under the hood” it has a lot of integrations hidden: chat with the operator, integration bus, CRM, navigation and other services.

Bot Mark is able to advise customers on device models and place orders – the conversion of users who have entered the product selection branch into a purchase is 5%. The user can completely create a shopping cart: select the device, its color, configuration, specify the payment method and delivery method.

4For the convenience of the user, a storlocator is integrated into the bot – the ability to determine the nearest points of sale and service by geolocation. The coordinates can be sent to the geoservice installed on the smartphone and build a route to the point in it.

Since the bot retains the ability to contact the support operator, for cases when the number of requests exceeds the number of support agents, we have developed a custom algorithm for queuing, which was not originally inside the platform.

Ilya Chernyakov

The co-founder of Cleverbots, developed the project from the very start

“One of the main advantages of support automation is the ability to collect up–to-date analytics on user requests and behavior. For example, now we can track exactly what the most popular queries are: order status, referral program, troubleshooting. A very small proportion of requests are “Ask another question”, which implies the connection of the operator. Our bot has outgrown the “bridge” stage between customers and operators and has become a full-fledged functional application inside the messenger.”

Vladimir Lazarenko

Customer Care Project Lead

“The main task that we wanted to solve was to give users the opportunity to easily and conveniently solve their problems and get answers to questions in a few steps in the bot, without waiting for the operator in the queue. To do this, after analyzing customer requests in text channels, we implemented scenarios for all the most popular requests, and now we continue to refine our chatbot, adding more and more new service scenarios.”