Welcome to the Listening Post!

Digital Research Renaissance

Tweet

 

At the MRA annual conference, Steve Alexander of Serendio and Michalis Michael of Digital-MR.com announced the coming "Digital Research Renaissance", proclaiming the "greatest research opportunity of the last 100 years is right in front of us." The speakers described four principles of digital market research:

1. Embrace Big Data - Originally data was scarce and had to be collected by researchers for there to be anything to analyze. "For years we struggled with getting customers to talk," Steve said. "Now they are talking about everything. The amount of raw data being generated on the web is staggering and has implications for MR, even if only 5% of the data is useful." Today there is a "data deluge" of external and internal data: customer email databases, call center records, CRM systems, community panels, MIS and ERP systems, web traffic data, ad network data. "Traditionally we think about sample populations because it is infeasible to get the entire population. With big data we can get everything. What to do with the data is more challenging." Rather than sampling, you can analyze the entirety of a data set, searching for patterns.
2. Adopt a network mindset - Viewing members of the population in isolation isn't enough - you want to understand their own influence and networks as well. For instance, Yahoo! research into instant messenger usage found that 20 million of its users had one friend, and one user had more friends than anyone: 30,000 friends. "This power-law distribution is very normal in a network where relationships matter: many have a few relationships; a few have many." Much remains to be learnt about social influence and idea propagation: "we are at the Wright brothers' stage."
3. Use a listening framework - A framework of Sources, Tools, Curation and Insight is needed. The human is essential. "This is about induction, not having a point of view, listening, framing downstream analysis as opposed to the standard research approach of framing a set of questions." Sentiment monitoring has evolved from counting (parsing), classifying (positive or negative using keyword sentiment) to categorizing (feature-level sentiment). "It takes two or three weeks to create a good taxonomy for analyzing [domain-specific] content harvested from the web."
4. Always connect the dots - "People think that you push a button and get data and that ends the conversation; the data you get starts the conversation." From neurons to Da Vinci, the human researcher ties it altogether. For instance, how does Share of Voice compare to market share? "One rule of thumb we've heard: If your share of voice is 10% higher than your market share, you should expect market share to grow by half a percent." Can you learn to predict changes in your market share based on your digital Share of Voice?

 The speakers shared three case studies of digital research projects:

• Anxiety treatments - They analyzed anxiety medications for overall sentiment and the primary sentiment drivers by category (efficacy in general and specific to anxiety, adverse effects in general and specific side effects). Lorazepam had the highest negative sentiment, and Ativan had the lowest negative sentiment; the sentiment for Prozac was primarily driven by perceptions of its efficacy, where sentiment for Ativan was driven by efficacy, drowsiness and addictiveness.
• Notebook computers - A Share of Voice analysis for notebook/tablet models estimated the importance of attributes by the number of posts mentioning them: the most important attributes were body color, battery life and CPU. Analyzing the models by sentiment score showed that most models had predominantly neutral sentiment, with the exceptions being a MacBook Pro model, the iPad and the Dell Studio 15. [Interestingly, given the debate over social media research ethics, the presenters showed quotes with people's real names (when available) and screen names with the comments they had said on sites like Amazon.com and Buzzillions.com.]
• Banking/Finance perceptions - A case study of finance and banking was heavily weighted towards Twitter, which had the majority of posts drawn from six sources. Share of Voice on Twitter was highest for Citibank (36%), then American Express (21%) and Bank of America (21%). Analyzing sentiment across the categories of bank, credit card, customer service and fees, there was extensive negative sentiment for Bank of America's customer service.

"There are digital ad agencies, there are digital PR firms, but no names of digital market research firms roll off the tongue, and that has to change," said Michalis. Questions researchers need to ask themselves about their organization:

• Who owns "digital"?
• What is my "digital" strategy?
• Am I spending enough?
• Am I spending it the right way?
• How do I choose my "digital" partners?

Michalis concluded, "We need to claim this space as researchers. We have a lot to offer. This is the era for research."