Attention, negative! AI monitors social networks

Attention, negative! AI monitors social networks



Monitoring of social networks has become a mandatory task of contact centers. Their employees are at the forefront of protecting the brand’s reputation. The future of the company sometimes depends on how quickly and efficiently they respond to comments and posts of dissatisfied customers. Innodata is trying to earn money by offering a platform for automating this process.


Most large Russian companies use social networks in one way or another to interact with the audience. As a rule, they hire employees whose duties include daily monitoring of social networks, primarily to work with negativity: detecting negative customer reviews and reacting to them. Such actions can prevent the “viral” spread of unwanted information, when a negative message is massively reposted, it gets into the media, acquiring new, sometimes not reliable, details.


A quick and high-quality response to a complaint allows you to change the client’s opinion about the company, and in some cases-even increase his loyalty. According to the American Sprout Social, 48% of users call the prompt response to a request the most important factor when communicating with a brand in social networks. According to a study by this company, in 2015-2016, the number of customer support requests through accounts and groups in social networks increased by 18%. Moreover, customers who need support first turn to social networks to find the company’s accounts.


Studies carried out by experts from the universities of Hatford and South Carolina have shown that customers have a close connection with the brand if the company works with the audience through social networks. Such customers are distinguished by increased loyalty and satisfaction with the quality of the goods and services offered.

The State of Social 2016 report showed that 54% of companies provide customer support through social networks.


Ivan Fedorov, head of the Yota public communications group, says that his company actively monitors brand mentions in the media space and its specialists are always ready to help users, whether it is a message in a personal account in social networks or on a thematic forum, website or media. To track brand mentions, Yota uses special monitoring systems, which allows you not only to see the mentions themselves, but also to configure the tool so that they are sorted by the tone of the message and topics based on keywords.

“We see any information that is not hidden by the user’s privacy settings, but even if the information is available only to the user’s friends, as is often the case on Facebook, there are always people who are ready to mention the brand in the comments and pay attention to a difficult situation,” says Fedorov. — As a result, almost no mention that requires an answer remains without our comment. The response time depends on many factors, but on average we try to respond within 1-2 hours on external sites.”

Other data



Innopolis resident Innodata Company presented a solution designed to work with customer requests, based on Big Data technologies and predictive analytics in social networks.

According to the company’s representatives, it can be applied in key sectors of the economy: retail, telecom, transport and logistics, the oil and gas sector, large E-Commerce, civil aviation and the financial sector.


The system is integrated and can exchange data with CRM platforms already installed in companies, such as Siebel, Microsoft or SAP. The data obtained in social media is transmitted to machine learning algorithms, which allows you to determine the category and tone of a particular mention in real time, helping to correctly prioritize responding to appeals in social media.


When working with requests, the system constantly analyzes fresh data and continuously self-learns to improve accuracy. Based on the accumulated historical data, the company has built mathematical models that allow, when receiving a request from a client through any communication channel, to determine the topic of the request with a high degree of probability, as well as the department responsible for processing and solving the problem. This approach has led to a decrease in the use of resources for analyzing and reassigning received requests.

According to the executive director of Innodata, Alexander Sergienko, the company tracks 95% or more of messages about the brand in all popular social networks and allows you to respond to a negative message with a delay of no more than a minute. The system will find the message even if the person did not directly contact the brand, that is, did not use the hashtag or the name of the official page of the company.


Sergienko told the publication Banki.ru that one of his company’s clients found out about the problems in his business before the local employees reported it. The manager saw a surge of negative comments in the system and learned about the accident.

In the future, Innodata plans to develop a chatbot that can communicate with customers automatically, without the participation of a live operator.

Go to our cases Get a free quote