Default Image

Months format

Show More Text

Load More

Related Posts Widget

Article Navigation

Contact Us Form


Sorry, the page you were looking for in this blog does not exist. Back Home

How Data Science Can Help With Negativity On Social Media

    Negative reviews on social networks, comments and generally negative activity associated with brands can harm the whole reputation of companies. Also, experienced PR-managers can not always predict what  one emotional story from a dissatisfied client will cost their brand. By not reacting in time and allowing the situation to get out of control a company can lose billions in just a few hours.

    Data Science For Social Meida

    KitKat and the loss of shares because of Indonesian orangutans

    On the 17th of March 2010 Greenpeace launched a media strike against Nestlé, namely the KitKat brand. The organization found out that Nestlé's Indonesian supplier was abusing deforestation to extract palm oil and that this was causing complex damage to the ecosystem. Greenpeace posted a video on YouTube that parodies the ad «Have a break, have a KitKat» and demonstrates that this «break» for office workers is built on the blood of Indonesian orangutans. 

    Nestlé reacted harshly, demanding that all videos be removed, citing copyright infringement. And while the community's reaction to the video itself was relatively restrained, Nestlé's reaction sparked real anger among social network users. Thousands of reposts and angry comments sprang up. The result: Nestlé’s share prices collapsed by 1.5% at the March 19th opening. The company lost about $2.5 billion in just 2 days, based on market capitalization.

    Honda and the sledgehammer a customer took to its reputation

    Another example - in 2004, Bangkok clothing store owner Duenphen Silaket bought a Honda CR-V. It was her dream car; she had been saving up for several years. However, almost immediately after purchase parts of the car began to break down one by one. Duenphen was visiting her mechanic almost daily. The car mechanic told her that out of 100 units of the CR-V model, one was bound to be defective, and Duenphen was just unlucky.

    An angry Duenphen turned to Honda's management for an apology and compensation, but they ignored her. That's when Duenphen took a sledgehammer and smashed the car to pieces right in the middle of Bangkok. Of course, she didn't forget to invite journalists to the event. The news spread through the media, becoming viral, and Honda lost 20% of its capitalization over the next three months. Duenphen was eventually compensated anyway, but the losses and the damaged reputation of the company had not been prevented.

    How Data Science helps you combat negativity?

    These examples show that it is necessary to fight against negativity in a timely manner, and that each reaction by the company should be carefully considered. There are always pros and cons, and in different situations the same answer could either extinguish negativity or cause it to grow exponentially. Data Science tools can help companies analyze the situation and develop an optimal strategy for responding to negativity. 

    For about 10 years, scientists have been analyzing negative comments on Facebook, Twitter and other social networks, and assessing the damage they have done to the world's largest companies. The researchers have also developed a number of recommendations for timely identification of such hot spots of negative comments and for limiting the damage they can cause. They are as follows: 

    #1. Identifying potentially viral posts

    About 3% of all negative posts on social networks become viral. Scientists managed to identify a number of general characteristics that such posts had:

    Emotional language. Users of social networks are more willing to comment and repost those publications that broadcast the emotions of the author. Such posts cause more sympathy and increase the desire to help.

    The way of communication. Usually the author is supported by the part of the community which shares his style of language.

    Overall: a dissatisfied client who writes emotional comments and knows how to find a common language with the community can be a real threat to the company's image.

    #2. Preventing the spread of negativity

    Negativity cannot be ignored - this is the most obvious pattern that scientists have identified in the research process. If a company does not respond to a negative comment that threatens to become viral, a loud scandal is almost inevitable (remember the case of the Honda CR-V and the response of its executives). 

    Let's highlight some of the most common types of reactions to negative comments, which helped to diffuse the situation and reduce the desire to engage with other users:

    ● A suggestion to discuss the situation in personal messages.

    ● Apologies

    ● Explaining why the situation that caused the negative occurred.

    ● Expression of sympathy.

    ● Offer of compensation.

    The first two reaction options are usually well-received and help prevent a scandal. However, the reaction to attempts to justify oneself, express sympathy and offer compensation may be unpredictable.

    For example, according to statistics, when a company starts to respond by offering compensation, it causes a wave of negativity and complaints from other clients who also want to get money. On the other hand, if the company does not offer compensation at all, it risks making a disgruntled client and his supporters angry. Therefore, the moment of offering compensation should be chosen wisely, when the first emotions have subsided and the client is ready for constructive negotiations.

    #3. Damage limitation

    Sometimes it is not possible to prevent the spread of negativity about a company, even by taking the best steps in terms of probability analysis. But even when a post with a client's claim has become viral, the damage can be limited, but it is necessary to act correctly for the situation.

    Trying to convert communication with a dissatisfied client into a personal dialogue is already dangerous - the situation has become public and the community will perceive this attempt as a desire to "hush it up quietly". Also, you should not just try to offer monetary compensation - it may be perceived the same way. The best option is to enter into a dialogue, openly discuss the situation and give detailed explanations about the incident that caused the negativity. And only offer money after this so that it does not look like a payoff. 


    Data Science allows you to offer new, more accurate and cheaper solutions to traditional problems. We at Andersen are actively developing the Data Science area and regularly share successful cases with solutions for our clients.

    For example, we used the following algorithm to identify negativity in the media:

    ● We compile lists of sources, keywords and tags.

    ● We write a cloud solution on Google, AWS, Azure or Yandex, which, using an API, periodically goes through popular social networks and target forums and downloads content to Data Lake from there. 

    ● Content passes through ETL Pipeline, is structured, cleaned and saved in Data Warehouse

    ● Semantic, lexical and tone analysis is done with structured data using the art models of natural language processing or provided by cloud providers of AI solutions. Mathematical modeling of the potential virality of posts and emotional response of the audience is also performed. 

    ● Information is provided to the customer in the form of dashboards based on ready-made BI systems (Tableau, Power BI, Data Lens, Data Studio, QuickSight) or on a custom-designed website.

    Other Articles:

    No comments:

    Post a Comment