What we're up to at BigVoice

Consumer Research Study: Social Media Mining
Expand or collapse case study

Mining Brand Switching Data and its Related Causes for Diabetes Drugs

Download PDF - Product Switching Case Study Download Product Switching Case Study

The Challenge

People often discuss their medical problems, circumstances and experiences in public discussion forums on the web. These archives of public social media forums contain valuable gems of useful consumer knowledge within volumes of useless junk. The challenge is to find them without incurring huge effort and cost.

This poses two important questions:

  • Can we use the Web and social media as a marketing research sample?
  • Can we quantify the rich, yet unstructured, information consumers post on the web?

Using the web and social media as a consumer research sample creates the following research opportunities:

  • A combination of observational and descriptive marketing research
  • Permits both qualitative and quantitative information analysis
  • Non‐invasive (no demand effect)
  • Minimizes recall error
  • Very rich data
  • Sample size is not an issue
  • Real time data

The web and social media also create the following challenges:

  • Massive amount of data
  • Data is all over the Web
  • Data is unstructured
  • Sample set is not necessarily representative

Our approach allows us to quickly and cost effectively address these challenges by using our proprietary software tools to search out and scan vast amounts of data, mine for our desired content, organize that content and create benchmarks of psychographic information. We can easily replicate these searches on a regular basis to monitor and update our findings.

Drug Switching Study Objective

Our objective for this project was to create a database of diabetes drug information from online forums that contain information on why patients stopped using a particular drug and the replacement drug. Additionally, we sought to better understand the consumer behaviors associated with diabetes drug usage and switching.

More specifically, we sought to answer the following questions by efficiently mining information from public medical forums:

  1. What diabetes medications people stopped using?
  2. The specific reasons they stopped using them.
  3. And, with which medications they replaced them.

Summary of Benefits

Using our Consumer Research Platform to mine public medical forums, we were able to:

  1. Quickly & cost effectively mine information from hundreds of thousands of respondents’ data about:
    • Brand switching
    • When respondent started and stopped
    • Reason for starting and stopping
    • Replacement drug
    • Reason for replacement
  2. Create a benchmark measure of psychographic information – who are the consumers?
  3. Develop an inexpensive yet robust, automatic ongoing database of information that allows you to track information over time and identify factors related to changes
  4. Leverage you consumer research budget for any type of product

Input Required

In order to develop this Consumer Research Program we simply required general directional information including:

Minimal key words & concepts:

  • What keywords are most associated with the brand and the industry
  • What are the concepts you are most interested in learning more about

Target forums and social media sites:

  • What online forums and social media sites most likely contain the information we want
  • Are there any specialized vertical industry forums?

Specific questions to be answered by public data:

  • Brands used
  • Reasons for brand switching
  • Replacement product
  • Reason for replacement

Our Approach/Solution

To build our database and capture psychographic insights, we used a patented software-based relation learning and extraction framework (CARE-II-HPSG/VC*) to mine public online medical forums. The software automation process included:

  • Building a list of known diabetes drug names by “scraping” drugs.com.
  • Creating a diabetes-specific lexicon by pattern learning from 20,000 sentences downloaded from the following forums:
    • Diabetes.blog.com
    • Diabetesdaily.com/forum
    • Diabetesforums.com/forum
    • Healthboards.com/boards
    • Forum.lowcarber.org
  • Clustering patterns identified and defining output relations (semi-automated)
  • Post-processing of complex relationships

We analyzed 650,000 forum messages consisting of approximately 300,000 non-duplicated sentences totaling 40MB of ASCII text. We found 2,732 distinct instances of diabetes drug replacement information.

The Results

  1. We were able to effectively identify the drugs with the most frequent usage stoppage.
  2. Create a list of the adverse drug reactions (ADRs) that were frequently mentioned as having a negative relationship with each of the diabetes drugs.
  3. Identify the most frequent replacements for each drug as well. The chart below actually shows the overall replacements regardless of the original drug used.


Based on a random sample of messages manually checked, our software was:

  • 98% accurate at identified usage-stopped diabetes drugs
  • 93% accurate identifying the replacement drug
  • 93% accurate identifying the reason for stoppage


Fast, simple to use, cost effective method of collecting valuable information from a large number of data points (650K). Can be used alone and/or with other types of research.

Goes beyond standard text mining and uses patented technology to transform and quantify the rich, yet unstructured, information consumers post on any and every social media platform

Search the web and social media for key consumer insights derived directly from consumer comments online combining the best of observational and descriptive marketing research.

Review and structure massive amounts of data to create a cogent picture of near real-time consumer sentiment that can be easily and cost effectively replicated.

We believe the efficiency and accuracy of this platform will change the way companies learn from consumers.

Sebastian Professional
Expand or collapse case study

Digital Look Book

Sebastian Professional Online Look Book

Click image to view screens from the site

Download PDF - Digital Lookbook Case Study Download Digital Lookbook Case Study


Sebastian Pro’s project goal was to gain publicity among influencers in the beauty industry for its crop of new star stylists and the new products featured in the stylist design creations.


To connect with influencers, BigVoice transformed the standard flat fashion LookBook into a user-directed non-linear experience.

“We were looking for something that would visually tell the Infinite Possibilities story using Sebastian Pro products. The LookBook that BigVoice created captured exactly that spirit, “said the DeVries Senior Account Executive on the project. “Not only did they come up with an amazing creative solution, we were incredibly impressed with the multiple times we came back to BigVoice with client changes they quickly delivered on.”


The response was tremendous. The outreach was targeted at writers at salon trade mags, influencer consumer mags like Paper, and Zinc, as well as star bloggers.

Every influencer from blog or magazine who was invited to the LookBook wrote on Sebastian Pro’s behalf.

Editors called the LookBook “inspiring” and embraced the playful way they were able to browse, compare styles, and download style recipes and product info.

Internally, the LookBook has become a reference for Sebastian Pro as a way to bring styles to life.

Unanticipated Benefits:

  • The LookBook gained traction as a resource passed from stylist to stylist and as a fashion reference for consumers to print out.
  • The Sebastian Pro sales force is currently using it extensively as part of its sales efforts.

To see more, visit the Sebastian Pro LookBook


P&G Bounce
Expand or collapse case study

Bounce Social Media Campaign

The Bounce Who-Da-Thunk-It Machine

Click image to view Bounce app intro

Dryer sheets? What’s there to talk about?!

Actually, we found lots of people talking. In fact, there’s a whole world of people who use Bounce sheets for purposes beyond the dryer. Moms stuff them inside the fingers of their kids’ hockey gloves to keep them from stinking. Scuba divers wipe them on the inside of their wet suits to make it easier to get in and out. Women wipe Bounce on their panty hose to keep their skirts from clinging in winter.

We created a social media campaign to help people share these ideas with others. At the center of the campaign was a Facebook application called the Who-Da-Thunk-It Machine.

Our goal was to make the experience truly social, to make the behavior mirror the ways people actually share ideas in real life. By building curation intelligence into the application, the user is able to share ideas only with individuals they think would find them useful. When a user submits an alternative use, the machine searches their social graph for friends who've indicated in their profile an interest in the subject. Their idea can then easily be shared with one person or 100.

By enabling users to share an idea only with friends who they think will find it relevant, Bounce avoids one of the most common social media pitfalls —unintentional spamming. Because what’s compelling to one friend, can be junk to another.

Expand or collapse case study

The first 3-D Home and Renovation Design Game

Inheritance Renovation and Design Facebook Game

Click image to view screens from game

Can you recall the manufacturer of the medicine cabinet or the vanity in your bathroom? Do you know the brand name of that cool anodized steel coffee table you bought? Walk the aisles of a Home Depot or any home furnishings store and you'll see hundreds and hundreds of brands that are a part of millions of consumers' homes. These are brands people purchase and use ever single day, yet they have no idea of the names of these brands.

So we created Inheritance, a Facebook 3D Renovation and Design game in which players renovate and design a beautiful aging mansion to their tastes. The game becomes a meeting place between consumers and a legion of formerly anonymous renovation, décor and furnishings brands.

Inheritance provides more than just placement, it enables actual interaction. Because it employs 3D modeling, the game offers a unique opportunity for consumers to actually see how products look and feel in a kitchen or dining room they're renovating. The net result is that when Inheritance players walk the aisles, once faceless brands, now have recognizable names (hey, I know Pegasus mirrors!). A relationship has been created.

Our investment in Inheritance represents our belief that all brands, no matter their size, can build a bigger voice with their consumers.

Inheritance is currently in beta. Stay tuned!

Beauty Cares
Expand or collapse case study

Spreading awareness of domestic violence

Beauty Cares video

Click image to watch "MakeUnder"

Beauty Cares is an organization that works with survivors of domestic violence in the shelter system. It's supported by a core donor base from the fashion and beauty industries.

The organization came to us to help generate awareness of its role in the survivor recovery process and help bring new thinking to the general discussion about domestic violence. Specifically, they asked us to develop a piece to communicate a difficult issue: When does behavior in a relationship cross the line into abuse?

Domestic violence is not a subject that people are necessarily comfortable with, or willing to share freely. And that was a potential obstacle to getting core donors, who were Makeunder's audience, to share.

So our idea and execution had to be powerful, but not make people turn away, or worse, tune the message out. We conceived and wrote "MakeUnder" and co-directed it with Sundance award-winning cinematographer, Dan Gold.

"MakeUnder" enabled Beauty Cares to make a powerful connection with its audience: Insiders in the beauty and fashion industry.

Additionally, the video's text-to-donate function enabled Beauty Cares to receive significant contributions from a wider audience as well.