In 2013, I was an embedded team member at IDEO during an externship swap. I was there to help IDEO think about how they could use data for prototyping during a project. To begin, the IDEO team performed some qualitative research to learn what consumers needed and wanted, then we ran a quantitative survey online. As the results trickled in, we started to analyze the data.
My jaw dropped
10+ different versions of Excel spreadsheets proliferated throughout the team of 4 designers. Our results needed to be analyzed quickly, but it was very difficult to keep track of what things meant and who had the most recent “clean” version due to the haphazard nature of the analysis. I begged the team, “Please, just give me your data and an afternoon and I think I can help.”
Later that day, I showed them an early version of what would eventually become catcorr.js, a JavaScript library that is useful for visualizing correlations between categorical variables. Being able to see all of the data in one place was an immediate win for the IDEO project team. We were now able to quickly and collaboratively explore different narratives from our qualitative research in order to evaluate whether or not the quantitative survey results supported or refuted our hypotheses.
Over the past few years, Datascope has worked on a couple of other projects that needed a visual comparison across several dimensions of categorical data, much like the analysis for that initial IDEO team. With this, we realized that there was a broader opportunity for catcorr.js beyond a simple visualization package.
An opportunity
You see, the default mode for analysis in consumer research today generally involves creating and printing out “crosstabs”, then scouring the results by hand to highlight significant and interesting findings. Crosstabs, which have been a mainstay in the industry for 40+ years, allow an analyst to “not miss anything”, but present an overwhelmingly large view of data and lack the flexibility, automation, and intelligence that have become pervasive in analytics tools across other industries. As a data scientist, I found crosstabs appalling and concerning. With dozens or even hundreds of rows of dense figures in a single table, how could you be sure that you wouldn’t miss a significant insight? How could you be sure that your personal biases weren’t influencing the story that you were telling about the data? How might we improve the process of consumer research to focusing on developing novel hypotheses instead of sweating the details of crosstabs?
From a v0.3.1 open-source package to the release of Elemetric
During the summer of 2016, Datascope launched a project to turn catcorr.js into something more compelling and robust that went beyond intuitive visualizations. We were interested in developing an automated insights engine that could quickly identify the most compelling results in an unbiased way. Yoke Peng and Damien did an awesome job and quickly got us to a minimum viable product (MVP) that we could put in front of customers.
At the same time, we began performing due diligence to size the market and better understand the tools, processes, and companies within the consumer research space. It turned out that our hunches about fragmented tool sets and processes were true. At a deeper level, though, we found underlying issues in the space that went beyond the use of fragmented analytics software, including a dependence on expensive and inefficient agencies, an expectation of projects to take multiple months to return results, and the confession that survey data was often inaccurate or poorly contextualized due to the pools of respondents being used.
Additionally, the industry was changing rapidly. When most market research platforms were being developed, data was an expensive luxury, brokered by firms like Nielsen or IRI that had achieved the massive scale required to provide relevant data. Modern companies don’t struggle with data–in fact they have more than enough–the biggest issues are around modeling and developing insights that can be leveraged in real-time, to meet increasingly rapid business demands. Ultimately, the problem turned out to be the same as it had been for decades, to convert a hairy business problem into a quantifiable survey, clean up the data, analyze the results, and tell a story to key stakeholders that makes sense and is actionable. We began to see clearly, though, that incumbent providers were doing an increasingly poor job providing solutions that fit the needs of a new generation of agile, data-first companies.
With an area so ripe for disruption, we realized that it would be foolish to create this product under the Datascope umbrella. The new concept needed to raise funding. It needed to be ruthlessly dedicated to product. And, most importantly, it needed to be laser focused on consumer research (at least to start). We needed to create a new entity, Elemetric, run by CEO Daniel Mason.
In the months since, Elemetric has made substantial progress toward realizing the vision of creating a consumer research platform for the smarter, faster needs of data-first consumer products companies. Elemetric hired Neel Kothari to lead the development efforts, and has honed its vision and extended the functionality of the original application. Elemetric is building a survey and research platform that sits on top of the breadth of information constantly flowing through marketing, sales, communication, and eCommerce software within consumer goods companies today, turning a company’s user base (customers and non-customers) into a powerful and inexpensive wealth of information to inform product and marketing decisions. Elemetric is working with selected companies in a closed Beta program, and and is currently participating in IDEO’s Startup in Residence program. Additionally, Elemetric is close to closing a pre-seed round funding round from angel investors.
We couldn’t be more excited to announce the launch of Elemetric, the goal of which is to help consumer researchers do what they do best: consumer research (not data munging in Excel).