The science - and scientist behind Thematic

Thematic was founded by Alyona Medelyan, who holds a Ph.D in Natural Language Processing and spent over 15 years researching ways to extract meaning from text. Her Ph.D at Waikato University’s renowned Machine Learning lab was sponsored by Google.

Alyona’s research covered the many ways algorithms can capture topics text, and resulted in the open-source software solution called Maui, which has been adopted by hundreds of researchers around the world. She has also published dozens of peer-reviewed articles that have been cited by more than 2000 academics.

Built to understand feedback

There are a multitude of Natural Language Processing solutions, released by the best teams at Google, Amazon, Microsoft and IBM. However, they were all designed to analyze generic documents, such as news articles. Applying these NLP APIs to customer or employee feedback results in disappointing results: a long list of keywords is extracted but most of the meaning is lost.

Thematic was built for customer feedback analysis, using the latest Deep Learning research and mimics the market research guidelines on how to create code frames from customer feedback:

Why Thematic?

Bottoms-up taxonomy generation

Thematic doesn’t presume what it needs to find in the data. It starts by identifying the most prominent themes in feedback.

Iterative code frame construction

Thematic uses semantic similarity to continuously create themes and construct a 2-level code frame from these themes.

Transparency and human feedback

While powerful, AI algorithms don't have implicit business knowledge. Thematic is built for anyone to easily teach our AI what matters to them.

Thematic Analysis versus Other Approaches

Other solutions in the market typically offer three kinds of approaches. These solutions take months to setup (apart from word clouds), Thematic provides insights in just minutes.

Approach
What it’s great for
Why it fails
Why Thematic is better
Word clouds
Visualizing answers to single-word questions, e.g. “Which word describes culture at company X?”
Word clouds lead to incorrect conclusions because they don’t group words and phrases into themes.
Thematic identifies themes in feedback regardless of which words were used to express them.
Taxonomies, text categorization & rule-based approaches
Analysis of feedback in business areas that don’t change, e.g. 'pricing'
Taxonomies miss the “unknown unknows” and are difficult to evolve over time.
Thematic uncovers themes you did not expect your customers to care about.
Topic modelling
Quick insight into sequences of related words in data, e.g. attributes people use to describe staff.
The results are hard to interpret and change.
Thematic’s results can be easily understood and modified based on context.

Get in touch to increase your key customer metrics today