3 steps to adopt online behavioral data

Can you name the 5 last websites you visited last night?

As a matter of fact, you can’t. 29% of the respondents don’t mention any of the websites observed using a “Passive Data Meter”, an app that tracks their online behavior. Another 28% just hit one website on their list. Yes, people perform poorly when remembering their online activities.

For a number of reasons, our memory is absolutely overwhelmed when recalling online behavior: increased range of events, reduction of time dedicated to activities, multitasking, and a low attention context make it extremely difficult for research participants to come up with an accurate recall. Gerald Zaltman, professor at Harvard Business School, said in his classic text book How Customers Think: “the correlation between stated intent and actual behavior is usually low and negative”, it turns out that the same principle applies to stated recall.  

According to the Q3-Q4 2017 Greenbook’s GRIT study, only 35% of Market Research (MR) suppliers are currently using Big Data Analytics, needed to deal with the sheer volume of data points that a passive data meter produces. Given that is hard to name a path to purchase in any category that has not been touched by digitalization, (from apparel to toasters, from cell phones to cars) brands from many industries could be badly missing the mark with their digital consumers. We have a big elephant in the market research room that is holding back its value creation capacity.

How to tackle digital consumers' behavior.

To track online behavior we basically have two approaches: a site-centric measurement, or a user-centric one.  

1. Site-centric measurement: When users browse a website they leave a track on the web server that they visited that could be analyzed and interpreted for research purposes. However, these “tracks” or “logs” are only accessible by the web server owner, and they also fall short when researchers are trying to understand who is behind them (many tracks are bots).
    
2. User-centric measurement: Users willing to participate in research can install an application (a passive data meter) that tracks their web-browsing and app usage for market research purposes, typically in exchange for incentives. Since those users are pre-identified, this methodology better addresses the “who” behind the “what”, and also provides a more comprehensive view of a user journey across different websites and devices. However, as it is impossible to force everyone to install the app, passive data metering has the intrinsic bias that any sample framework has.
   
   

 Site-centric research has been well adopted by many companies for a while now, given that they typically needed it to manage their online presence and customize their web products and contents. Thus, some MR end-clients are well-positioned to deal with the technical struggles that Big Data Analytics poses.

Nevertheless, when it comes to analyzing competitors and market positioning, the user-centric measurement is the tool of choice, and a passive meter is needed along with MR suppliers that can make sense of this data. Today, the paradox is that not all suppliers have had the opportunity to deal with big data sets, so they lack the technical expertise. There is hope, though. The good news is that an online behavioral data ecosystem is already in place for MR suppliers, that includes behavioral data collectors, tools for analyzing big data, and data scientists or big data analysts.

 In order to adopt online behavioral data, take the following steps:

• Select the right behavioral data collection supplier: Data privacy matters. Under the GDPR, Personally Identifiable Information (PII) is defined as information that can be used to distinguish or trace the identity of an individual, either alone or when combined with other personal or identifying information. So, PII is not only your email or home address, a track of your mobile phone geoloc could become PII too. Making sure that participants have agreed to be tracked by giving clear and explicit consent is mandatory. As the Facebook / Cambridge Analytica scandal showed, these consents cannot be buried down in “terms and conditions” or legal jargon. Be inquisitive with your behavioral data collector and do your due diligence. Hint: if there is no opt-in panel you may have a problem. When you see companies with “black boxes” or not willing to disclose how they collect their data, it usually means consent is non-explicit.
  
  
• Hire talent and learn the tools: Just with non-mobile data, a panel of 200 participants can create around 1 million rows per month,- yes, big data is not just a buzzword, your Excel cannot handle this. To analyze these big data sets, programming statistical languages like R or Python are needed, so hop in and ride. Another key aspect is understanding the relevance of taxonomies. Analyzing a collection of URLs or apps can be overwhelming without a proper taxonomy, so don’t overlook the tagging and categorization process required. Remember that one size doesn't fit all, given your research purposes you may need to establish a particular taxonomy (e.g. Do I include Uber in the “travel” category? What about GasBuddy?)

• Don’t get trapped in the middle with your research briefing: Either have very specific questions or embrace data discovery but never, ever, fall in the middle. For instance: “which websites are more popular for mothers-to-be and babies’ mothers?” This is a specific question that behavioral data can tackle with accuracy. Or “what online touchpoints drive purchases in the spirits category?” Again, online behavioral data is key, but in this case, an exploration journey is needed. You will only be able to know what data is actually useful after you have dived deep into it. Curiosity is your most valuable tool in this case.  

And finally, don’t forget the “why”.

We cannot rely on people’s responses to understand digital consumers. Market research practitioners need to face this new reality and adopt online behavioral data. Nevertheless, for the right issues, asking questions still is a powerful tool. Don’t forget that the “why” won’t come merely from observational data. A unique combination of survey and online behavioral data will help you to unlock greater value.

One of the hardest things for an incumbent company to do is to disrupt itself, your human instincts shout against it. Remember that disruption represents an opportunity and bright lights too, so move resources before is too late or you will be disrupted by someone else.

Trust the process and reinvent yourself.