Civis Data: The powerful dataset built right into Civis Platform

October 31, 2017 Wade J.

I work alongside data scientists every day, so I know that data science teams can help their businesses thrive if they have the right tools and the best, most complete data. That’s why I’m happy to share that Civis Platform comes with a powerful dataset built right in. With accurate, predictive data included, Civis Data helps organizations learn more about current and potential customers so they can see the bigger picture and make better decisions.

Businesses need a full understanding of their customers to make smart decisions: Which customers will churn? Which messages work best, and for whom? How should sales staff be allocated to win more business?

To get that full understanding of their customers, businesses need two kinds of data: the data their own business generates, and data from outside their walls. Data science teams have a choice: they can just use their own data and hope for the best; they can cobble together a solution using public or licensed consumer files built for marketers (which often have thousands of messy columns that offer a lot of noise and not a lot of predictive value); or they can trust other data scientists to build tools they can use with confidence.

We know this struggle firsthand because we’ve been there: When Civis Analytics started, we primarily did data science consulting work and encountered this incomplete or inconsistent data problem frequently, so we built our own suite of tools to solve some of the fundamental issues data scientists face as they scale up. Here’s how we built Civis Data to solve the problem:

  1. We started with sources similar to what others use to build their consumer insights, but we didn’t stop at building a simple list with demographic characteristics.
  2. We made this data ready for machine learning by trimming uninformative values that add noise (more on the how and why here), filling in missing values, and incorporating reliably predictive public data from sources like the US Census Bureau.
  3. We used the latest in open-source machine learning to add best-in-class models for race, sub-ethnicity, marriage, and more. Civis data scientists did this using over 500,000 survey responses (we do market research too).
  4. We used Scripts and Workflows in Civis Platform to automate the build and quality control of Civis Data. This brought visibility and repeatability to every step of our build process. Having quality control metrics at the beginning, middle, and end of the process gave us confidence that the file we’re delivering is the best quality.
  5. And lastly, security and privacy are extremely important, so we decided to deliver our data in Civis Platform, which meets the highest, most stringent level of third-party validation in enterprise security — SOC2 Type II certification — allowing our users to keep personally identifiable information separate from their analytics workflow.

A client of ours in the sharing economy needed to put their own data into context. They had reams of data: data on how often their users interacted with their brand, time in app, satisfaction scores, and other unique survey data. But they did not have an ability to analyze their customer base through demographic or behavioral measures. By appending consumer features from Civis Data to their customer list, the company was able to build an acquisition campaign based on which potential customers looked the most like their best existing customers. With these scores in hand, our customer was able to activate their list via social, online, and direct mail campaigns and reduce their costs by more than 20%.

Other customers have used Civis Data to:

  • Bring context and understanding to customer or donor data with demographic and behavioral indicators.
  • Append predictive features to their customer database to build machine learning models for customer or donor acquisition.
  • Build location-based models for store analysis to decide where a new store should go, or how to cluster store types for marketing and assortment.

At Civis, we do our best work and empower our customers to do the same when we can open our toolkit and be fully confident we have the right data for the job. With Civis Data built right into Civis Platform, we’re excited our customers can share that experience every time they open their laptops.

The post Civis Data: The powerful dataset built right into Civis Platform appeared first on Civis Analytics.

Previous Article
Civis R&D Bookshelf: AlphaGo Zero, testing machine learning code, and more
Civis R&D Bookshelf: AlphaGo Zero, testing machine learning code, and more

This post is part of our Bookshelf series organized by the Data Science R&D department at Civis Analytics. ...

Next Article
Civis R&D Bookshelf: AI Research, Claude Shannon, and Software Abstractions
Civis R&D Bookshelf: AI Research, Claude Shannon, and Software Abstractions

This post is part of our Bookshelf series organized by the Data Science R&D department at Civis Analytics. ...