The right research
tl;dr - At Google I created a script that analyzes large amounts of text and shows relevant stats. This lets other writers quickly analyze text for research, auditing, and more.
Background
Waze is a driving app that lets drivers tell other drivers about hazards on the road, like traffic jams, crashes, construction and more.
When it came time to redesign how drivers report these road hazards, we wanted to validate and explore the terminology around these reports.
But how could we tell if the labels we wanted to use were actually how people speak about traffic and traffic-related incidents?
Research
After going to the usual sources like Google Trends, landscape analysis, and past iterations, I found myself listening again to radio traffic reports from my hometown of Chicago.
I briefly thought about transcribing the broadcast until I remembered the Internet Archive had a whole catalogue of already transcribed radio broadcasts.
Solution
I created a python script to analyze the transcripts for frequency data. Analyzing tens of thousands of words in the traffic reports, I was able to understand the terminology used in our key markets.
Aside from raw data, I also created word cloud visualizations (like the picture below) so people could quickly understand the results.
Impact of the project:
Besides ensuring the right terminology is being used, this tool allows other UX writers to quickly gain insights into large amounts of text and use it in their everyday work.
I also integrated this tool into my larger scorecard tool as well.