What got that length of time in their head? I’m sure if someone at a cocktail party came up to them and asked about a new product, the answer would be either far shorter (if the questioner looked like they were just making small talk) or much longer (if the answerer sized up the questioner as a potential customer).

Shouldn’t the conversation start something more like: We’re looking to make this specific point to this specific audience…what do we need to say, and how? (Notice that this question is neither medium or length specific?)

Now, the modern TV commercial is based on divisions of a minute. But that’s because it was an easy unit for the provider of the medium/channel to define and sell. In other words, ABC isn’t in the least bit interested in how long it takes to get your point across. They define the parameters—because they own the channel—and companies figure out ways to buy the appropriate amount of time and shoehorn (or stretch) their content to fit. But online, there’s no such constraint.*

So if the medium isn’t defining the length, why do people still tend to default to length as the primary variable of a piece to be created? I think it’s because they are ascribing traits to their audience. The marketer is assuming they know how long it takes for the average target consumer to “get” the information…or they are assuming they understand their target’s attention span. Either way, it’s an assumption made without good reason.

This is just one example of where we go wrong when demographics or psychographics of an audience are assumed. Marketers often don’t really know or understand their audience at all. So they have to apply some simpler attribute to frame what “success” looks like for the piece. It’s about as far as you can get from thinking strategically.

The same approach applies when you start thinking about getting into social media. Some companies start with a basic idea that is bereft of tangible objectives: “We want to make a Facebook fan page.” But have you first taken the time time to know whether your audience is active of Facebook? What is the ultimate objective of creating a fan page and how it will support your overarching marketing strategy? (There are brands, believe it or not, that might be spending their time and money better in places other than social media channels.)

And there’s an equivalent in market research endeavors. “We need to build a research community comprised of our consumers!” But have you evaluated the successes of other brands with communities, understand how it can apply to yours…and will it support the organization’s learning needs as well and as efficiently as other available tools?

In any case, remember to understand the audience before defining the piece. Remember to think strategy before tactics.

*Yes, YouTube will cap you at 15:00 minutes. But if you’ve got more to say than that large time allotment, you can make a series.

PHOTO CREDIT: http://www.flickr.com/photos/karynsig/ / CC BY-NC-SA 2.0

Mobile platforms offer a number of advantages in terms of providing the ability to capture in-the-moment feedback and shopper insights, particularly when geolocation information is integrated into the mix. And, as Joel Rubinson points out, the speed at which marketers are embracing and leveraging mobile platforms threaten to leave market researchers perpetually “two steps behind”.

The advantages of mobile market research are obvious. The platform is by its very nature embedded in the lives of consumers. This serves several important functions:

First, we are so tightly tethered to our mobile devices and access them so frequently, that I suspect properly designed research touchpoints will be considered less intrusive than many current efforts (more on this below).

Secondly, the portability of mobile devices provides a variety of opportunities to better understand consumers in their native habitat. Shopper insights could be taken to a whole new level and provide much better understanding about decisions made at the shelf.

In addition, incorporating geolocation/social networking elements provides another dimension of valuable data via platforms that are engaging consumers as opposed to disrupting or annoying them.

Lastly, the evolution of touch interfaces will enable collection of data from consumers via an experience that is far more interactive and engaging than traditional online surveys.  The net result is better response rates and, potentially, improved data quality.

The challenges of mobile marketing research are just as obvious. Namely: relatively limited distribution in the marketplace, widely varied development platforms, and the difficulties of delivering surveys that gather nuanced data when in situ. Let’s start with the lack of distribution. Although current penetration rates for smart phones are still relatively low, this number will continue to rise.

• Current Cellular Phone Penetration: 91%
• Current Smart Phone Penetration: 17%
• Projected Smart Phone Penetration by end of 2010: 33%
• Projected Smart Phone Penetration by end of 2011: 50%

However, even with increased adoption rates, the market share of individual brands or devices is not sufficiently large to support broad research initiatives. It seems the primary barrier to further expansion of research is technology. Specifically, the development of apps/platforms that can provide access spanning the highly splintered mobile market. To point out how daunting this challenge is, consider the relatively simple comparison of the app development process for the iPhone vs. Blackberry provided by the Wall Street Journal. While the iPhone process is relatively straightforward, the proliferation of models provided by Blackberry and the resulting variety of screen sizes, features, and processing power make app development an incredibly complex task. And that’s for Blackberry alone.

Josh Mendelsohn’s recent piece touched upon the importance of creating “mobile friendly” surveys. The first part of that equation is keeping surveys short, simple, and to the point. The second aspect is making sure the surveys are easily accessible and navigable on mobile devices. And while there are some platforms that currently support mobile research, there is still a long way to go. It’s certainly an area of need and it’s simply a question of when, not if, this capability will be developed.

Mobile platforms have already demonstrated strong potential, as outlined in the Marriott case study, also detailed in a most recent issue of Quirks. The most exciting piece will be the ability to mesh geolocation data, sources of retail information (bar codes, geolocation, etc.), and other pertinent pieces of data with research responses. The methodological possibilities have barely scratched the surface. The only question is how quickly the technical issues above can be addressed and allow the learning curve to be scaled.

PHOTO CREDIT: http://www.flickr.com/photos/mathplourde/4639156283/ / CC BY-NC-SA 2.0

I think it might look a lot like this:

Yes, I know there’s been plenty of posts and articles proclaiming the Old Spice “Reponds” campaign both an incredible success, and others proclaiming it a raging failure. Either way, I think it needs to be viewed as equal parts “marketing” and “market research.”

Here’s a description of the methodology for choosing which tweets, comments and posts to respond to:
“In the room there are two social media guys and a tech guy who built a system pulling in comments from around the web all together in real time. We’re looking at who’s written those comments, what their influence is and what comments have the most potential for helping us create new content. The social media guys and script writers are collaborating to make that call in real time. We have people shooting and we’re editing it as it happens. Then the social media guys are looking at how to get that back out around the web…in real time.”

The videos aren’t being posted in chronological order immediately after the Tweets and comments they are in reply to. They get moved up and down a queue in a deliberate, orchestrated, if very fast way.

— Iain Tait, Global Interactive Creative Director at Wieden+Kennedy

At the core of what’s described is an immediate feedback loop. So as comments are made, videos are crafted and delivered. Which spawns more comments. Which means you can see which messages (videos) spawn the most views and comments. And because those comments are made on social media sites like Twitter or Facebook, you can learn at least a little bit about who’s making the comment—their gender, location, job, age, number of connections (i.e. “influence) and many more details are all available. It’s a real-time focus group… a means to optimize messages on the fly to better appeal to specific demographics.

Compare that to sales data—collected via loyalty cards, surveys, etc.—in the months following the campaign and what have you got? You’ve got the ability to see exactly what demographic is responding to your specific marketing messages to the point of becoming buyers. That’s some serious market research power.

Granted, not every brand or product is going to be able to pull this off. You have to have a fairly well developed brand personality that targets a specific demographic or consumer group that is highly involved in social media. Creators and Critics, in particular, to use the Forrester descriptions. It just so happens the group Old Spice is targeting fits both desciptions. And they have been honing this point-of-view and voice for several years now.*

So… is this the future of social media based market research? I think there’s a good chance it fis.

*Here’s my still-favorite ad from the new vein of Old Spice marketing. It’s from three years ago:

As I eagerly dug into the piece, a short summary of findings from a recent study by 360i, I eagerly anticipated the numbers that would validate my skepticism regarding Twitter’s utility for market research and fly in the face of many hard-to-believe claims regarding the number of people discussing brands on Twitter.  Initially, I was not disappointed as the article asserts that very few tweets (12%) even mention brands.  And though the article was primarily focused on brand marketing implications, surely the rest of the story would paint an equally morose picture for market researchers?  Yes…and no.

So what do the numbers mean?  How fruitful (or fruitless) is Twitter mining?  Below is a table that outlines some of the key findings from the study.

With over 65 million entries per day, the volume is impressive and daunting.  However, only 90% of those 65 million are considered “consumer tweets”.  Then, when one considers that only 12% of all tweets even contain brand mentions, it’s clear that you’re dealing with just a fraction of the available content.  Further, Twitter brand mentions are heavily concentrated in the categories of Social Networks (22%), Entertainment (17%) and Technology (17%), with Apple, Twitter, and Google representing approximately 70% of the 1,800 tweets sampled (as an aside, this is a surprisingly small sample size given the abundance of content).  However, even if your brand represents only 0.2% of brand mentions, you would still have an impressive 14,040 tweets a day to comb through.  And if you aggregated unique tweets over the course of a month, you’d have a significant base of content to analyze.

In the end, I was left with a few key takeaways…and yet more questions:

1)      The volume of content waiting to be analyzed is obscene.  This is generally a good thing, because even though it might seem like drinking from a fire hydrant….

2)      There’s a relatively small subset of brands out there (currently) that are sufficiently relevant to generate a significant base of content.  Which raises the question: what is the actual distribution of brand mentions on Twitter?  Obviously brands like Apple and Google have an embarrassment of riches.  But how many brands are sufficiently relevant to generate a reliable stream of content?  25? 50? 100? Whatever the number, it’s a tiny fraction of the brand universe.

3)      What is the distribution of tweets by user.  It’s pretty well-documented and intuitive that a large portion of Twitter content is produced by a small volume of disproportionately active users (this data more than a year old, so if there are newer/better statistics available, please feel to chime in).  If this usage pattern still holds true, what are the implications for analyzing the content? How large is the pool of users generating the brand mentions?  Are you oversampling a very vocal minority?  Motrin anyone?

4)      What is the quality of the brand mention content?  Specifically, what portion of the 12% of tweets actually provides relevant, usable information?  Anyone who’s ever coded open ended responses knows of what I speak.  I’m having a Bud Light right now. It’s not my normal beer of choice, but its cold and I like beer.  But if I tweet, “having a bud light and pretending to be smart”, does that provide any context or insight beyond a brand mention?  Perhaps if I tweeted, “Having a cold Coors Light. In blue mountains I trust…Eureka!” it would have a bit more value.  Suffice to say that the number of tweets providing context, tone, or some other valuable piece of information is much less than 12%.  A somewhat tangential point, but worth noting.

In the end it seems that consumers are having “conversations” on Twitter, but far fewer of these of these exchanges are brand-focused than the conventional wisdom might suggest.  And while there are certainly other relevant areas of exploration beyond  brand-specific contexts, it still seems that the range of current applications for market research is relatively narrow.  As with all other queries regarding the value of Social Media research, the answer is, “It depends”.

PHOTO CREDIT: http://www.flickr.com/photos/wiselywoven/3110939912/ / CC BY-NC-SA 2.0

Neither of these things reflect reality. These are examples of perception bias.

If you think honestly—really honestly—about how many times you are in line (at the grocery, fast-food restaurant, gate at the ballpark and so on) and how often it actually takes longer than anyone around you, the percentage simply isn’t that great. (This hasn’t stopped people from thinking they’re overwhelmed by Murphy’s Law and writing pointers on how to avoid it.)

So I have another hypothesis now that I’ve seen the light: People don’t hate marketing. (It’s become en vogue to say people do, but it’s not true.) People think they hate marketing because they are more apt to remember—and complain about—the times they are interrupted by marketing and advertisements that they don’t want want to hear.

The interesting thing about that new hypothesis is that experiments to prove it have already been done. A lot of them. Because people continue to be influenced by marketing and advertising, spending money on the things that they need and want. Whether they saw it on TV, heard about it on the radio, were told about a product by a friend, or learned about a brand on Facebook, they were influenced and happily took action.

The marketing that people hate is usually disruptive for one of the following reasons: A. Marketing in the wrong place – it’s the right message interrupting the wrong audience; or B. Not knowing your audience – it’s the wrong message interrupting the right audience. In either case, the disruptive nature of the ad is conspicuous and triggers the perception bias/negative association with marketing.

(There’s also: C. It’s the right message in front of the right audience… but there’s some other folks seeing it who don’t care for it, too. Don’t worry so much about this one. This can never be completely eliminated, and collateral hate usually doesn’t become detrimental…even though some companies tend to think it might and they over-react quickly.)

The companies that best influence are also the ones who know their customers best. They’ve figured out that marketing and advertising in the places that target customers frequent, in a tone that is most appealing to those same customers, works. They’ve achieved success by monitoring, researching and testing cleverly crafted messages. Then they can get the right content in front of the right people at the right time. A pretty novel approach, huh?

Part 3 – The Way Forward

Social Media Research is already a viable and a valuable source of consumer insights, particularly for brands that are in high involvement categories and have sufficient penetration among those actively engaged in Social Media. The purpose of this post is not to question the potential of Social Media for market research or minimize the efforts made by those who have advanced Social Media research to its current position. These advances are crucial and should be applauded. However, there seems to be a lack of frank discussion regarding the relatively narrow band of predominantly qualitative applications of SM Research, and what it can and cannot do. There are still several challenges that must be overcome to unlock the full potential of these platforms and to begin thinking about the displacement of traditional market research tools.

As discussed above, qualitative research has traditionally been observational in nature while quantitative research has relied upon the practice of evoking responses from subjects with prescribed questions and stimuli. To what extent can we overcome the paradigm shift to completely observational techniques? Is this even a realistic goal? If not, it seems there is a need for advances in methodologies that will facilitate more “context-defined” data acquisition. Whether that is something similar to current survey methodologies or something completely new, there is a need to better define context. Those who are already active in Social Media will certainly demand higher levels of engagement, so it seems obvious that the solution will not be in the form laborious surveys with lengthy matrix questions and the like. And perhaps framing this question in the context of the traditional survey concept is itself a barrier to identifying better, more context defined methodologies? Regardless of what the the solutions are, moving beyond the purely observational is critical for the advancement of SM research.

Advances in text analytics and sentiment analysis are essential. The volume of data available is simply too great to rely upon human analysis, and the current state of text analytics/sentiment analysis is well short of what is needed in terms of accuracy and reliability. This is a formidable challenge, but advances will be needed in order for SM research to take the next step.

Solutions to the current sampling challenges will also need to emerge. Tracking measures are needed to better understand sample sources, background information, and other key elements needed to provide the necessary respondent context. This may be simpler fix than the challenges above, assuming the concerns over privacy can be adequately addressed. In recent HBR post, David Armano touched upon the sustained momentum of Blippy despite recent privacy concerns, indicating consumers’ increasing willingness to share data as part of a larger behavioral shift. However, the recent hullabaloo about Facebook’s privacy policy indicate that this is a sensitive topic and we are not yet to the point where all or even most consumers feel comfortable “passively sharing” personal information. Jeffrey Henning spoke to these challenges and the need to develop privacy guidelines for SM research at the recent CASRO Technology conference.

Fortunately, the trends toward increased sharing/less privacy and increased social media adoption will help to address some of the sample bias and respondent context issues. Thus, addressing the current limitations of the purely observational nature of SM Research and advancing text analytics/sentiment analysis would likely trigger the demise of traditional market research that has been prematurely declared. For now, if I’m a brand or product manager for an entity that is relevant in social media, it is imperative that I monitor the space as a responsible steward of the brand and mine relevant content to supplement my research initiatives. However, I would not feel comfortable relying solely or even predominantly on Social Media for business-critical insights. I would have far too many reservations about the rigor and projectability of the data to make key decisions for the brand.

PHOTO CREDIT: http://www.flickr.com/photos/anemoneprojectors/ / CC BY-NC-SA 2.0

Challenges

The nature of social media offers a unique opportunity for market researchers, but also presents some significant challenges. In many ways, it is not unlike the advent of online research. Initially there were many concerns about online penetration rates, obtaining representative samples, validating sample, and many other methodological challenges. Several of those problems linger today, although few would argue the viability of online research. Just as the interwebs opened a new channel for researchers, the conversational nature of Social Media and the opportunities for observing organic dialogue are unprecedented. However, the whole concept of Social Media is relatively new, and just as it took a while for researchers to figure out how to best leverage the internet for market research, it will also take researchers some time to bring order and rigor to the vast sea of social media data. A few of the challenges include:

Observational Methodologies, Context, and Automated Analysis

One could argue that the hurdles presented by Social Media are greater than those currently (or originally) presented by standard online research. This is partially because the nature of social media is far more conducive to observational (qualitative) research than interactive research (surveys, etc.). If we think about a continuum of research frameworks spanning from purely observational to purely interactive, Social Media research falls at the observational end of the spectrum, even more so than ethnography and other techniques in which subjects at least know they’re being observed and often have an idea of the topic of interest.

This doesn’t mean that the qualitative initiatives supported by SM do not have a place, it simply means that there needs to be an understanding about the current limitations. Tim Phillips shared this lament in his recent article, saying, “In fact the increasingly social manner in which technology is used has brought about a shift in research towards qual – not quant.” Thompson Morrison also opines that SM research is dominantly qualitative and should be applied with caution.

While the conversational nature of social media provides many opportunities, it also forgoes control for context. It requires making assumptions or inferring meaning in many cases, and generally limits the ability to extract specific information or precisely frame context. While automated tools have proven relatively effective for analyzing SM data to track public opinion on simple questions (Candidate A vs. Candidate B), there are significant concerns regarding the accuracy/reliability of automated tools used for sentiment analysis and the like. MROCs (Market Research Online Communities) represent a happy medium of sorts, offering the ability to facilitate/host organic conversations (admittedly this is not purely observational) while combining the ability to guide conversations, ask specific questions and elicit responses to stimuli. In addition, MROCs provide a known, profiled sample base. Which brings us to the challenges associated with Social Media sample.

Sample

We are just now approaching the point where internet penetration rates and panel management practices are sufficient to support “reliable” samples. This is usually in the form of convenience samples, as anything approaching a “representative sample” (a hotly debated topic in and of itself) will be such a time-consuming and expensive proposition that the costs outweigh the incremental benefits. The bottom line is that there’s been an ongoing discussion regarding the reliability of online samples for many years, and rightfully so. In light of this, how can we assume reliable samples from SM research? Aside from Facebook, penetration rates are still very low. They’re lower still if you focus on the pool of regular content contributors. So sample is not only drawn from a small minority, you’re sampling the vocal portion of the small minority.

Another challenge related to sample is context.  Unless you’re provided access to the background variables by the network or via a third party, you have disturbingly little context. So your choices are to manually assign the appropriate context (assuming that information is available) in terms of demographics (much less purchase behaviors, attitudes, etc.), or to have your trusty automated tool infer this information. In the former case, you’re looking at quite a bit of manual effort that would translate into significant costs for even a qualitative sample size. In the latter, you’re relying on a machine to do something that even experienced professionals would have difficultly doing with a high degree of accuracy.

For the same reasons outlined above, data reliability, projectability, and overall data quality are also major hurdles. These elements are already under intense scrutiny for online panels that are much more tightly controlled and managed than Social Media platforms, so one can only imagine the statistical pitfalls of sampling the social graph. Concerns over the reliability and projectability of data acquired solely through Social Media channels will continue to linger. The current situation seems to be more analogous to the Wild West in comparison to the relatively controlled world on online panels.

PHOTO CREDIT: http://www.flickr.com/photos/shortformvideo/ / CC BY-NC-SA 2.0