Man vs Machine

So, Big Data. The market research industry continues to struggle with the concept. It was one of the buzzwords of 2013. Some have come up with a big data offering. Some are searching for a point of view on it. Some counter with small data. Many still have only a vague sense of what we are talking about.

I have asked many colleagues and clients what this concept means to them, in the hopes of developing a brilliant solution that would make me wildly successful. Well, this seems to be taking some time, but anyway, I’d like to share with you what I have learned so far. As I am working in the healthcare sector, this is my focus below.

1. Big Data (in Pharma) is IMS data

For some of my pharmaceutical clients, all they can think of when asked about large data sets is IMS data. IMS captures and sells information about the prescribing behaviour of physicians at the pharmacy level. Through this data, pharmaceutical companies track the sales of their products.

2. Big Data (in Hospitals) is Patient Records and Interaction Statistics

Healthcare providers, particularly hospitals and other large organizations, capture myriads of data on patients flowing through the system. The analysis of this data is largely off the radar screen of traditional market research, and falls under the discipline of health informatics.

3. Big Data is Social Media data

This is a view that many market researchers adopted when social media first appeared on our professional horizon as another form of human expression. Last year’s MRIA NET Gain conference, dedicated to big data, featured a number of presentations in this area.

For those who do not want to develop their own proprietary solutions, subscription-based social media analysis tools are available and used by both end clients and market research vendors.

4. Big Data analysis is a different way of saying Data Mining

Some sectors have worked with large data sets for some time. I am thinking of scanner data in retail, and loyalty programs (Air Miles, Petro Points etc.). Fifteen or so years ago, the statistical techniques used to sift through such data sets were called ‘data mining’.

This practice is still ongoing, and the size of data sets ever increasing with more and more customer touch points being added. Some think of this type of analysis, when hearing the words big data.

5. Big Data is Data that is created by Machines

This type of big data is rarely mentioned and obviously not in the forefront of a market researcher’s mind. However, it has grown exponentially and is increasingly viewed and used as a source of customer information.

For market researchers, the question (and the fear) is to what extent human analysts are still needed, and to what extent ‘the machine’ can do it on its own. And how we can make sure we are still needed.

We say: “You need an expert to interpret what your data means.” We say: “A consultant is needed to guide the analysis process.” We say: “Meaningful data analysis is the development and testing of hypotheses, and only people can come up with those.”

And we are right.

How many times have I looked at the results of a statistical analysis and said, “This does not make any sense.” And then we discarded the analysis and started fresh, because results need to make sense. To another human. To your client. They need to lead to actionable insights and recommendations.

So we are still needed. But…

But many, many processes are now automated, from data analysis over producing charts even to highlighting key insights in charts. Far fewer people are needed to work with the data then before. Take a look at www.beyondcore.com – will get you thinking.

Some companies function with very little market research, in the sense of interactions between real live researchers with real live respondents. Machine-generated user data that streams back from devices guides the refinement of these products. New Apps are developed by split testing and seeing how early users interact with certain features, following the logic of clicks. Service companies integrate their customer interface and CRM software with their enterprise management system. Automated cues let managers at different levels know how they are performing, and notify them if there is a problem.

Technical skills are essential for survival. How can you tell a successful agency these days? If you look at their ‘careers’ page, most of the open positions are for developers (i.e. IT people). Those who win in providing business intelligence are either companies who are focused on the digitalization and automation of data collection and analysis, or companies who make intelligent use of available software products and platforms within the research process.

Do you know what Hadoop is? A wireframe? CSS? If not, perhaps it is time to google it right now…

Man vs Machine

 

The Role of Motivation in Patient Engagement

As Dave Chase @chasedave recently stated The most important medical instrument is communication, and Patient Engagement is the Blockbuster ‘drug’ of the century. The idea is that by engaging patients to become more proactively involved in the management of their own health, better outcomes can be achieved, and generally at a lower cost to the healthcare system (compared to expensive tests, procedures and medicines).

New technologies make it easier to engage patients – for example via online portals, health apps, personal electronic health records, portable monitoring devices or a clever combination of these tools within a new care model. Clearly, when the generation of texting, tweeting, vining, instagraming twenty-year olds turns fifty and start their decent into chronic illness, resistance to digital, mobile, sharable health tracking and communications technology will no longer be an issue.

Remains the issue of motivation. The people who continue to supersize their burgers and fries, do they lack awareness of the health impacts of excess weight? The schizophrenic who skips her pill because it makes her head feel fuzzy, has she not been educated on the dangers of messing with her medication regimen? Will improved communication with a healthcare provider convince my uncle to stop smoking?

Yes, there are those who suffer from inertia, who are uncertain about the right way forward or who find it hard to fit taking care of themselves into their busy days. These patients will find it helpful to be supplied with tools and supportive healthcare providers who make it easier for them to look after their own health. These patients are the low-hanging fruit for the new care models.

And then there are other people. People who will not download the app. Who will not sign up for the e-newsletter. Who do not want to be called by their pharmacy to remind them to refill their prescription. Some people will continue to do dangerous, unhealthy things because they want to. It makes them feel good, at least momentarily. Some do not want to face the realities of getting older, of their failing bodies, loss of beauty and loss of agility. Some are comfortable with the thought that this is all inevitable, and do not feel inclined to take action. Some are looking to their doctor for the quick fix, just make it go away, I don’t want to bother with it.

PtEngageUSED

Motivating people is a tricky business and tech tools are only going to do part of the work. What motivates patients to take care of themselves? Pain? The desire to live longer? A feeling of obligation? Because your mom told you to? Peer pressure? Because it is cool? Fun? Because it makes you look better?

Also, doesn’t motivation change over the course of one’s life? What motivates a twenty-five year old to track his weight loss with a health app and what motivates a seventy-year old to continue his androgen deprivation therapy would likely be very different things. Capturing the Show Me The Way segment of patients with new patient engagement tools will be easy and rewarding. Addressing the Maybe Later and the No Way segments will be much tougher, and cracking the motivation nut will be essential to make it work.

Sea Change in Market Research

The art and skill of market research lies in asking the right questions and drawing the right conclusions from the answers. I know how to ask questions.

I know how to ask them online, on the phone, in person, in a fashion that makes responses quantifiable, in a fashion that allow us to publish the results, in a fashion that elicits emotions, in a fashion that minimizes bias, in a fashion that entertains my clients. I know how to ask questions to old people, to young people, to people with illnesses, to people with children, to CEOs, to large donors, to physicians, to nurses, etc. etc.

In a house, with a mouse, in a box, with a fox, here and there, I can ask questions anywhere…

What if market research is no longer about asking direct questions to real, live people? Why are we asking questions anyway? Our clients want to know what people think and feel, and what they will do, based on their thoughts and feelings. How they will vote, who they will support, what they will buy.

Much of this can be elicited from data that is produced without asking questions. I recently read an article on how you can predict someone’s age, gender, sexual orientation, level of education and the emotional state he or she is in relatively accurately from the pattern of likes they leave across the Internet. Predictive modelling is the name of the game. How can you link likes, content of posts, tweets and comments to action, online and offline? The best people who develop these algorithms sit no longer in traditional market research companies.

They sit in large IT companies. Or they sit in smaller digital shops, where they specialize in a particular thing. And probably in some large financial institutions. And government think tanks.

What do they understand about people? What do they not understand about people? What do my clients need me for? Sure, I know my clients business. I consult. I interpret. I put things in context. At the end of the day, it is still all about making the right connections. So you know what pattern of online behaviour precedes a purchase. Now what? What information do you really need, and how do you use this information to your advantage? That is where the consultant comes in.

To do the job right, however, the consultant needs to understand what kind of information is out there, what is technically possible, what is practical and what is economically feasible in terms of analysis. And to stay on top of that is becoming more and more time consuming with the data explosion in which we are currently caught up…

Innovation, the new normal?

Companies in many sectors are facing rapid change. Following Clayton Christensen’s terminology, established businesses are being disrupted by new technology, and new business models are developed around these technologies.

Whether it is 3D printing of medical implants, crowd sourcing of clinical trial data analysis, software that supports pre-clinical studies and identifies the most promising drug candidates, ‘big data’ capturing patients’ genomic profile or personalized health records that patients can carry from physician to physician, fundamental transformations are afoot in the healthcare industry.

Consultants to the healthcare sector struggle to stay on top of all the different angles that are emerging. How much reading can you do in a day? Should you rather update your skills in data mining (i.e. working with ‘big data’) or become an expert in social media platforms and the many ways they are being used by patients and physicians or study government initiatives to incorporate new technologies in reorganizing the way healthcare is delivered to the patient?

The state of confusion is pretty typical for market changes. Initially, there is a whirlwind of new ideas and approaches. Are electric cars going to be the way of the future or ceramic fuel cells? Or will biking emerge as ‘disruptive technology’ in a reorganized urban neighbourhood? Will patients carry their own health records around on a USB stick or will they become universally accessible through a (password protected) cloud? Will pharmaceutical companies find ways to make drug development cheaper or will fewer drugs be approved or will best supportive care with the bells and whistles of comfortable retirement living ‘disrupt’ the oncology pipeline? Will iOS, android or Windows 8 emerge as the dominant ecosystem for computer / tablet / phone or do we need to learn all three to know what works how in which environment? Etc, etc.

Should we wait until the dust settles before we decide how to focus our efforts?

I am not sure that the dust will ever settle. The pace of change is accelerating, with no sign of stopping or settling down. Then where should we pitch our tent? What should we hold on to? I believe that companies and individuals will succeed who develop mechanisms, routines, practices that allow them to deal with change. Not just once, but on an ongoing basis. Those who effectively survey what is going on in the whirlwind, who systematically capture their own ideas on how to ride the storm and who devise an easy process that allows them to test, develop and implement these ideas will have a chance.

Is big data transforming healthcare marketing research?

Yesterday (June 27, 2013), SAS and GSK announced a collaboration which puts clinical trial data ‘in the cloud’ in a secure way, respecting the privacy of trial subjects, and makes it accessible to other researchers. It is believed that other big pharma companies may follow suit and create an unprecedented shared data base that could potentially speed up the analysis process, make analysis more transparent and produce significant advances in medical discovery.

While this particular example of ‘big data’ pertains to clinical trials, many other big data sets exist (or are being created) in the healthcare space, awaiting data integration and analysis. One wonders to what extent this trend will impact the need for primary healthcare marketing research. Secondary data analysis is not new – it has been part of business intelligence for a long time. What’s new is the amount of data that is being collected, the multitude of platforms and interfaces through which it is collected and the ease with which the data can be accessed and analyzed.

Traditionally, primary marketing research has been faster than secondary data sources at delivering behavioural data such as prescribing of certain drugs. This is now changing. Mobile health apps, EMR, data warehouses for adverse event reporting, point-of sales data at the pharmacy level and many more points of data collection are becoming more readily accessible. Secondary data is going ‘real time’, well almost. On the other hand, primary data collection methods can also harness the power of ‘real time’, thinking of mobile surveys etc. Who will come out on top, or rather, which mix of primary and secondary sources will deliver the best insights?

Also, primary marketing research has been practically the only way to capture attitudes and beliefs and to explore how they relate to behaviour. Arguably, communicating with your target audience is still the best way to understand their motivations. However, social listening, drawing on tens of thousands of online conversations and powerful tools for text analysis, has made some inroads into this area as well. In addition, to what extent do stated opinions really drive behaviour, and how good is primary market research, even with creative methods and advanced analytics, at uncovering these drivers?

Big data is certainly transforming the primary marketing research industry, in healthcare as well as in other sectors. The question remains which solutions will bring the most value to clients and will become the new standard for companies who survive the transformation.