Varis for Social Media Monitoring
Leena is a marketing professional at a large consumer food manufacturer. She is responsible for following social media channels (Facebook, Twitter and WWW forums) to find out consumers' opinions.
Her responsibilities are:
- notice changes and trends in consumer opinions
- respond to important messages
- produce interesting content that gets likes and shares
- internally report potential product quality problems to production
- internally report promising product development ideas to business
Before Varis, her challenges were:
- reading all messages is a lot of manual and repetitive work
- important messages may be accidentally missed
- each social media channel has its own user interface, creating a fragmented experience
- off-the-shelf social media monitoring tools have the following issues:
- difficult to customize
- poor Finnish language support (an important segment for the company)
- no understanding of own or competitors' brand names and their relationships
With Varis, Leena's day looks like the following:
Leena opens the custom Varis application that provides a unified view into all social media channels.
She first checks for messages that require rapid action. Varis has custom alerts for messages that indicate unhappy customers, and for potentially offending messages. Leena responds to unhappy customers and reports quality issues internally. Varis alerts are trained using real data and are therefore relevant.
Leena got a question from the product manager of Rainbow Candy, a recently introduced sweets product. The manager in interested in the overall consumer feeling of the product. A problem Leena has had with off-the-shelf social media products is that they have difficulty distinguishing the noun rainbow from the brand name. Varis, however, understands custom brand names, so Leena can quickly locate messages related to the product. Varis estimates each message for positive or negative sentiment, so Leena gets an overview of the consumer feeling and can zoom in to messages with strong sentiment.
Leena wishes to write Facebook content related to desserts. She first searches for recent dessert-related messages to see current trends. She types 'dessert' into the Varis smart search box, and gets results related to cakes, ice creams and pies. Varis has a semantic understanding of words and thus knows that cake is a dessert.
Next, Leena drafts a Facebook message about desserts. To fine-tune the message, she uses the Predict Likes feature of Varis to approximately estimate how popular the message would be, based on its textual content. The predictor is trained using historical data of the Facebook page. After changing a few wordings, she sends the message.