T-Mobile

The Task

• To analyze of telco data from social networks for 12 months.
• To analyze campaign Volavka and its impact on social networks.
• To analyze campaign Bezpečnost and its impact on social networks.

The Result

T-Mobile gained insight into the social networks and impact of the marketing campaign. By comparing competition in time we have found that O2 customers are very dissatisfied with cancelling of their youth plan. Based on the analysis, the main topic for marketing campaign have been identified (e.g. availability of mobile network in different areas). Customers do not understand the term shared internet => budget reallocated to different campaign goals.

T-Mobile

The Task

• To analyze customer complaints and feedbacks.

The Result

Clustering of text messages done on monthly basis. SentiSquare is identifying the main negative and positive topics across all touchpoints (and customer segments).

O₂ CZ

The Task

• To mine plain text notes from business representatives and operators from call center.

The Result

Besides the most important keywords and phrases, SentiSquare also extracted leads from the notes in plain text. Our algorithms were able to identify customers who are ready to leave, but they are not in the retention process yet. Important was also identification of foreigners who do not speak Czech language. The core of our analysis is always to find the main topics of the analyzed texts, which we did.

O₂ Slovakia

The Task

• To analyze gigabytes of URLs.
• To segment clients based on this classification.

The Result

Enrichment of customer‘s profile in CRM. URLs visited by customers were categorized and thus created segments of customers according to their leisure time interests.

Nestlé

The Task

• Market research for a life-style magazine promoting health.
• 15 million posts from relevant sources to be analyzed.

The Result

Identification of top 5 topics that are most relevant for the market (= for the customer). Recommendations for articles based on season and topic attractiveness. Discovery of controversial topics (e.g. dukan diet) and recommendation to promote these topics. Identification of key words for PPC campaign (e.g. UM = umělé mléko; in English: AM = artificial milk). Quantification of language that customers truly use.

Liberty Global

The Task

• To analyze entertainment industry.
• To create TV series ranking.

The Result

SentiSquare created ranking, recommended which TV series to bring to Czech market + whether dubbed or in original sound. Important result was a recommendation for targeted ads of hairdresser studios - based on viewer's likes of famous actress's new haircut in TV Show.

Hewlett Packard Enterprise

The Task

• To analyze discussions of CTOs.
• To find how important terms Azure and Cloud are for CTOs.
• To find how CTOs think about implementation of HP platform.

The Result

Discovery of specific obstacles in HP client's minds - why they are reluctant to implement the HP platform. Cloud Business is the main topic of these discussions, Azure means a great deal to CTOs.

Česká spořitelna

The Task

• To discover root causes, moments of truth and efforts in customer complaints.

The Result

SentiSquare is cable of identifying root causes and moments of truth automatically. Moreover we can identify possibility of escalation to regulator before it actually happens. It was also possible to teach our algorithms to recognize the effort. A major recommendation for Česká spořitelna was stated: to improve processes that are being constantly criticized and thus remove the bottleneck.

A Bank

The Task

• To analyze millions of comments from internet.
• To discover the main topics.

The Result

Automatic clustering and classification of posts was performed, thus the main topics were discovered. But based on the bank's request, also the other approach was used - sorting out the customer posts into categories that were predefined by the bank. Managers in the bank were provided with access to semantic search engine - it displays not only the specific keyword but also other forms of it, and even post where the keyword was not mentioned but it is relevant to that query (search based on meanings, not just keywords).

ČSOB

The Task

• To analyze relevant sources for the whole fiscal year and find opportunities and threats to the bank.
• To find important moods and facts.

The Result

SentiSquare found out that comparison of banks is performed mostly in February and in the autumn and that mortgages are being discussed the most in summer - June and July. On certain sources ČSOB brand is perceived much more negatively - this is an important fact for ads targeting and for the particular communication messages. Another important insight for campaign management is on which online sources people complain more than elsewhere. Last but not least, SentiSquare found influencers and described their opinions, attitudes and language.

EY

The Task

• To analyze more than 500 reports related to ROP (Regional Operational Programme).
• To find reports that include the project impacts.

The Result

SentiSquare had to split up all pdf reports into thematic paragraphs, it was the key to find the impacts and identify the reports that contain them. On the input there was a big load of unstructured data (reports of various lengths) and the analysis transformed this into structured thematic paragraphs, in which it is easy and fast to do search, follow trends or perform any other tasks. Besides that the client received access to SentiSquare Analytics which is now ready to process any reports from civil service.