The Thematically Discover platform quickly identifies reliable and meaningful thematic categories in any kind of text response — open ended survey questions, focus groups, in-depth interviews, product reviews, and more.
WHAT WE DO
The research industry faces a tough choice when it comes to finding meaning in text:
machines aren’t trustworthy enough and people aren’t fast enough.
THE NEED TO CHOOSE.
Our central premise is that the best way to use AI is to accelerate human intelligence, not to replace it.
The key to the Thematically Discover platform is a unique feedback loop in which machine-generated insights are informed and refined by the analyst, enabling rapid discovery of themes in unstructured text. Machine intelligence is powered by a sophisticated computational model that discovers themes based on the underlying relationships among words and phrases in a document, not just keywords or frequency counts. A user-friendly interface puts the analyst in the driver’s seat at all times, letting the user direct the system toward a better understanding of their data.
WHY WE’RE DIFFERENT
10X FASTER ANALYSIS
A beta client — a large research company — had previously analyzed a set of forty in-depth interviews from an international project, which took about 50 hours. Starting from scratch, an analyst using the Thematically Discover platform needed only 5 hours to produce an analysis, achieving the same degree of meaningful insight.
In a Government Accountability Office study of Federal Aviation Administration staffing and workforce issues, GAO analysts took over 100 hours to manually produce a thematic coding structure from the transcripts of 12 focus groups, and then code the data into those themes. In our academic study using the same dataset to validate the Thematically Discover methodology, a Thematically analyst recreated a virtually identical set of themes from the same data and created a first-pass coding into the themes in under 10 hours.
Campus Compact, a Boston-based nonprofit, needed to analyze action plans (comprising thousands of pieces of text) from nearly 70 universities, to distill a set of categories capturing the needs and priorities of effective campus civic action. Without Thematically’s efficient, AI-driven approach, the cost of assembling this information manually would have prevented the project from being done.
Andrew Stavisky is a qualitative research guru. Philip Resnik is an AI natural language processing expert. When they met at a conference a few years ago, it was a chocolate meets peanut butter moment.
Andrew deeply understood the pain that unstructured text causes, and Philip saw how to build technology to take the pain away.
So we prototyped it. We validated it. We refined it.
Now, our Thematically Discover platform is helping clients across multiple industries get the insights they need.