MarketTools Blog

There’s been a lot of buzz lately surrounding Path to Purchase – the course of events or activities that consumers follow from the first glimmer of interest in a product up through the point where they actually make the payment to complete the buy. By understanding the components of the Path to Purchase, marketers are able to strategize ways to maximize a product’s impact upon the consumer at each phase to ultimately win the sale.

Like many aspects of market research, there’s a lot more to this process than meets the eye – it can be much more complex than you’d expect. While investigating your product’s Path to Purchase, you’ll uncover some interesting shopper insights along the way, such as:

Much of the Path to Purchase may purely mental. The actual physical path through a store from the shelf to the cash register can be an important aspect of your customer’s Path to Purchase, and it’s worthwhile to understand ways to overcome barriers and facilitate their way.  However, the most valuable insights often lie in the decision-making steps that precede the retail transaction.

Paths may begin long before the consumer ever reaches the store. Depending on the category, the consumer may actively seek out information beforehand. In fact, this research may dictate the specific channel or retailer they visit to shop for and ultimately buy the product.

Paths often vary wildly across categories. For large purchases such as cars, the path may start with a consumer’s dissatisfaction with their current car, leading to extensive time spent researching models online. For a pack of gum, the path may start with the sight of a flashy package near the cash register and end just seconds later.

Some of the insights you could expect to uncover by mapping out your customers’ path to purchase include:

•  Is your product a “destination buy” that gives shoppers a reason to leave the house and go to a store, or is it an impulse buy they throw in the cart when they’re already there? If the latter is true, perhaps you should think about ways to make your product more of a trip driver.
• Which other products tend to be purchased in the same basket as your product? This could lead to some great cross-promotion ideas.
• What do consumers like/dislike about shopping for your product? Perhaps there is a way to improve your customers’ shopping experience. Can you break down any barriers?
• What other insights can you uncover to assist retailers in selling your product in store? Providing insights to the retailer might win you some more distribution and shelf space.

Over the next couple of weeks I’ll be following up this discussion of Path to Purchase with questionnaire tips to help you effectively survey your consumer audience so you can better understand each phase of the path (Part 2 of the series), and tips on analyzing and interpreting your path-to-purchase data for maximum impact (Part 3)  Be sure to check back!

This blog post revisits some practical tips from from Jolinda Decad, one of our MarketTools CustomerSat Research Consultants, who offered up great advice on survey design best practices for Voice of the Customer programs in a 10-minute interview that you can listen to here.

Here are some of Jolinda's key considerations to keep in mind when you develop your customer feedback surveys:

• Focus on a specific purpose: As you begin development, resist the urge to pull together a cross-functional team to start brainstorming survey questions. The best surveys focus on a specific purpose and ask only questions that are relevant and actionable. The more people involved, the more likely the purpose will be clouded and the more difficult it will be to get the information you need from the survey.
• Clarify survey objectives: What problems are you trying to solve by asking customers about their experience? Often companies are looking to understand how to meet their customers’ evolving needs within a particular touchpoint, and it’s helpful to think through the end-to-end customer experience within that touchpoint. For example, if you’re evaluating customer satisfaction in your contact center, you’ll want to think through the typical contact center experience and determine questions you could ask about each aspect of the experience, such as how long they are on hold, how they are greeted, how their issue was handled, if there was any follow-up, etc.
• Anticipate action: Think through how you might incorporate the survey results into decisions and actions. This will help you in two ways:
   
  First, it helps you determine the demographic variables you need for decision-making. If you want to make decisions based on customer segment, region, product lines, etc., you can ensure you include these demographic variables in the survey to more easily segment the data for analysis.
   
  Secondly, you can ensure that questions are asked in a way that drives clear action without setting false expectations. For example, if you sense that some customers want longer support hours but need a better idea of the number of customers that actually require this, you will want to ask whether the current support hours are meeting their needs without asking a question like “would you use support after-hours if it were available?” Not only is this question leading the respondent, it also sets the expectation that you’re considering extending support hours, which you may determine isn’t necessary if only a small percentage of customers feel their needs are not being met.

Check out the interview with Jolinda to learn more about how to design a great customer feedback survey from one of our EFM best practice consultants.

And as always, MarketTools offers ready access to experts in the fields of survey design, customer loyalty, and market research who can provide advice about best practices.  Let us know how we can help you!
 

I’ve been reading Daniel Kahneman’s new book Thinking, Fast and Slow, and I highly recommend it to any market researcher.  Be warned, though: the implications can be kind of scary.  One of the topics the book covers is the “conjunction errors” that people commonly make – as demonstrated by the classic experiment known as the “Linda Experiment” (see page 156 of Kahneman's book). 

Linda is described (keep in mind that the experiment is very old) as “31 years old, single, outspoken and very bright. As a student, she was deeply concerned with issues of discrimination and social justice and also participated in antinuclear demonstrations.”  Participants in the experiment are asked to indicate which is more probable – that Linda is a bank teller, or that Linda is a bank teller AND active in the feminist movement.  Even experimental subjects who were highly trained in statistics tended to choose the conjunctive description (bank teller AND feminist) as more probable than the simpler, more general description, even though it is impossible for a conjunctive to be more probable than any of the components of the conjunction.

This can be easily seen by the Venn diagram, above.  The probability of being a bank teller AND a feminist can be at most equal to the probability of being a bank teller and that only occurs if all bank tellers are also feminists.

So, what does this have to do with Segmentation Profiling? 

The issue to explore is why people consistently make the conjunctive error in assessing the likely probability of events.  The answer, as Kahneman describes in experiment after experiment, is that once a plausible scenario has been constructed, all information about underlying base rates and probabilities is routinely ignored by the human mind.  By putting together various “facts” describing Linda, it is easier for the mind to find examples of people who fit those facts and attribute all of the other characteristics of those people to the Linda description.  An internal system that makes judgments by matching patterns replaces a system that is capable of more complex reasoning.

Now think about your last segmentation study.  Did the research company supply you with a handy “description” of each segment?  Of course they did.  Descriptions help make the segment “come alive” and make it seem more “accessible”. We find it easier to connect with a segment if we “know” that it is “more likely to be female, age 45-55, high income, and works for a large company”.  As demonstrated in the conjunctive error statements, this description is likely to be representative of a vanishingly small proportion of the actual people in the segment.  (“More likely to be female” is often concluded when the proportion of females in a segment is 55% instead of the 50% in the overall population). If we write ad copy targeted at this description, chances are good it will not be relevant to a large proportion of our target.

What can we do?  The tendency to simplify a description and ignore the actual results is hardwired into our brains and our clients’ brains.  To minimize the risk of losing this dangerous game we need to be aware AND make clients aware of the basic problem.  We need to constantly test our conclusions and recommendations against the knowledge that we are likely to assume a lot more about our segment targets than is supported by the data.
 

A recent blog post by Brandon Ellse on Research World’s RW Connect blog argues :

"How can behavioural economics practically change the market research landscape? This remains less than clear to me. We’ve all heard about the experiments, their control and treatment groups beautifully poised in anticipation of the 'ta-da' moment where 'real' human behaviour is revealed like a rabbit pulled from a hat. These behavioural economics-inspired tales seem to waft over our intellectual senses, leaving us with a strange feeling of certainty that only the perception of knowledge exclusivity can bring. I won’t lie, I’m impressed. Getting at causality is no mean feat and I am often left in awe of the ingenuity of these researchers, but I want practical application and I get the feeling that you do too."

The benefit of behavioral economics is the insight it gives into the proper specification of the models we use every day in marketing research.  Behavioral economic theories about how and why people overstate their estimates of frequency of purchase enable us to construct models that better predict actual behavior from the imperfect survey answers we collect.  Understanding that losses loom larger than gains enables us to construct appropriate non-linear price functions in discrete choice models.  

For example, behavioral economics tells us that threshold prices exist, that is, prices where exceeding them causes an unexpectedly steep decline in the price-demand function.  Knowing this has enabled MarketTools to specify pricing DCM models with threshold points identified.  I gave a presentation (Wildner R., and Conklin M., “Price Thresholds and Prospect Theory”, The 14th Annual Advanced Research Techniques Forum of the American Marketing Association) showing that pricing models with thresholds always predicted better than models without.  Using behavioral economics theories allowed the specification of a better, more predictive model.

The linear models of the past have worked the best when the true form of the model was close to linear. This provides a bias towards small incremental changes because small changes are going to be approximately linear.  To understand the potential for large gains or losses we need to have the right underlying functional form and behavioral economic theories help us get there.
 

Guest blogger Candy Michael is Vice President of Customer Feedback at American General Life Companies, one of the top insurance providers in North America – and a longtime MarketTools CustomerSat customer.

This past year, American General Life Companies launched a Voice of Producer (VOP) program series, to ensure that we deliver an exceptional customer experience for our distribution partners.  We leveraged our EFM system with MarketTools CustomerSat to maximize the success of our partner feedback program.

Our distribution partners are independent marketing organizations, insurance agencies and producers offering our products and services, plus those of our competitors.  Prior to our Voice of Producer program series, their opinions and satisfaction levels were filtered or anecdotal.  At times we heard that we were not easy to do business with in areas such as marketing tools, technology, and new business processing.

In today’s highly competitive world, we must know our distribution partners by understanding their needs, finding out what works for them and what doesn’t so we can respond.  The VOP program series is designed to survey American General’s distribution partners to:

• Determine overall satisfaction with us, so we can identify areas of dissatisfaction, pursue initiatives to increase satisfaction in various business areas and objectively measure results
• Gather input on specific distribution areas of focus to better differentiate our products and services from our competition
• Seek feedback on select products, services and transactions, which is usable for more finite measurements of service delivery and quickly identifying areas requiring service recovery activities.

The VOP program series design and development resulted from a collaboration between our Customer Feedback Office and key operational and internal marketing groups who served as a steering committee for soliciting feedback from distributors.  At one point we collected 11 topics and 200 questions from our committee!  Knowing that our distribution partners receive emails and surveys from many carriers, we did not want to frustrate respondents with long surveys covering multiple topics.  The outcome is that our VOP program series offers distinct surveys on specialized topics, covering Marketing Software and Advanced Sales tools, Underwriting and Issue, and Technology.

Averaging an 11% response rate, we leveraged MarketTools CustomerSat to conduct email-based surveys and send post-communication emails to 40,000+ respondents outlining their feedback and our action plan.  As a result, our business groups made modifications to business processes, including significant performance gains on software load times and an encrypted email attachment process.  All three VOP program series utilized segmentation within distribution upload records and branching, and skip patterns within CustomerSat to ensure that only relevant questions were displayed to appropriate respondents.  This same segmentation and question logic provided deep and relevant analytics, and comprehensive action management capabilities for each VOP program series. 

Our VOP Program series will be enhanced and expanded next year.  This year’s launch has given us a baseline measure of our distribution partners’ satisfaction and loyalty, in addition to establishing a service benchmark for key operational services.  Our EFM system, powered by MarketTools CustomerSat, has enabled us to provide powerful data collection and analytics to our organization to help us gain new insights about our partners in a cost-effective manner.
 

This is part of an ongoing series of blog posts on Mobile Market Research technologies from the MarketTools Market Research team.

Back in September I blogged about 5 things that every researcher must know about QR (Quick-Response) codes – those digital barcodes that allow users with a camera phone equipped with the correct reader application to scan the code and launch a web page, a survey, or otherwise connect to information. Since then, the hype around the technology has continued, countered by a lot of nay-sayers pointing out the limitations of QR codes.

Let’s look at some of these arguments against the technology and reevaluate if QR codes can still be an effective market research tool.

1.    QR code adoption in the US continues to lag Europe and Asia.  Some even say that the QR code “fad” has already come and gone.  Maybe it’s true that the technology will never catch on in the US as much as it has in other markets. There is plenty of evidence out there that Americans are less likely to adopt technology that doesn’t benefit us immediately (see: metric system). However, that’s not to say QR codes can’t still be a useful tool. Thankfully, the cost to implement QR codes as a facet of your research plan can be done with absolutely minimal cost. And the key word is facet. Obviously you wouldn’t make it the focus of your research plan. We’re not going to dissolve online survey panels or eliminate phone or mail research and only use QR codes. Printing a QR code on a product or displaying one in a captive area where you’re looking for specific feedback is a low cost way to collect on-demand, in the moment customer data. At the very least, for many of us, it certainly trumps typing in a long url on a mobile device.

2.    College students (a group that by all accounts should embrace QR codes) are still confounded by them. A recent study reported that the majority of college students with smartphones have no idea how to scan a QR code. This is a troubling statistic, but has been completely corroborated by MarketTools’ experience with this demographic. The key is to leverage QR technology in the right way so you can use the numbers to your advantage. Let’s assume that QR code access rate is only 3% among college students. If you’re trying to use QR codes to survey a particular college classroom, this obviously spells disaster! But what if you’re using a QR code to launch a survey of the readership of a nationwide publication targeting college students, or for a CPG product targeted toward college students – with an attractive incentive? All of a sudden, that 3% access rate yields you more completes than you know what to do with.

3.    Newer technologies will surpass QR code technology. In fact, Google has recently started phasing out QR codes to focus on a developing technology called Near Field Communication (NFC). NFC works much like a QR code, but is activated by proximity to a touch point rather than by scanning. In other words, you only have to hold your phone up to something rather than sift through your apps to open up a reader, hold your hand steady, and scan a code. What’s not to love?

First and foremost, it will be several years before NFC technology is included in US phones. Second, as I rub my research crystal ball, I predict NFC will suffer from at least several of these issues as adoption builds:

• Some mobile device manufacturers will drag their feet on including the technology while initial consumer demand is low
• Many consumers with the technology will struggle to understand how to use it, such as how to toggle it on/off
• The technology will come under heavy scrutiny by Congress as consumer privacy continues to come to the fore
• Some industry experts (like QR code “traditionalists”) will mock NFC for its low adoption rate and argue the technology will never work as a research tool

Did you notice that some of these barriers are the same faced by QR code technology? The point is that no new research medium is perfect – it’s a matter of carefully selecting the right mix of media that best meets your needs.

As fun as it is to make blanket statements about how a new technology is either the best thing since sliced bread or completely useless, the best approach is to acknowledge its strengths and weaknesses and think carefully about how to benefit from them. QR codes are not going to completely revolutionize the market research industry, but they will be around for a while.  They will be used successfully by those who understand how to take best advantage of the strengths of the technology, and make it worthwhile for the consumer to engage with them.
 

Every year ends with dozens of prediction pieces looking back on the past year or ahead to the next.  Heading into 2012, some of the “Top 10” lists are expanding to 12 to celebrate, with topics ranging from beer to mobile infrastructure.  While there are many year-end lists for the market research world, I’ll offer a single theme instead – Convergence.

Mathematically, convergence may signal the arrival of computational limits or even the irrelevance of sequence order.  However, in market research  “convergence” is more like what happened in the convergence of the telecommunications industry – where several services are available from the “same pipe”.  Here are some examples of areas in market research where we might see convergence in the year ahead:

• Mobile and Shopper research will collide.  A National Retail Federation poll recently found 41% of members were increasing investment in mobile retail and marketing.  It’s crucial to continue to develop and test new mobile research programs to help us continue to progress.
• Mobile is no longer the “3rd Screen”.  Many households are moving to digital entertainment and away from traditional TV, and the use of mobile streaming apps is on the rise.  For brands, the ability to track advertising, image and sentiment will get more complex and the sheer volume of data will require creative new models to make sense of it.  This is an area where research analysts can bring expertise to the conversation.
• Gamification of research will be extended to mobile devices.  While there is still much investigation into the best implementations for game-based research, the movement to mobile devices will create pressure on game-like and game-based research.  This will affect the validation of both gamification and mobile research.
• Economic issues in Europe will force brands to more closely manage pricing.  The interconnected economies of North America and Europe create the need to have a global aspect for domestic price plans.
• Pricing and Packaging will converge into a single study architecture.  The tradeoffs between changing or holding price points will drive new studies that collect attitudes on simultaneous changes to price, package size and quantities that are all variables for pricing strategy.
• Social Media drives statistical research, and vice versa.  Social media monitoring is being used to inform sentiment analysis across brands – however the elements needed to make predictive assessments are still missing.  Research analysts will increasingly experiment with adding social media metrics into time-tested models.
• Brands will begin using traditional methods to test and quantify social media trends.  As the flexibility and speed of traditional studies increases, it’s harder to ignore the value of social media data in helping us better understand consumers.  But what we need is the discipline of traditional MR methods to deliver predictive data necessary for business change.  This will be especially useful in proving ROI for event sponsorships as well as understanding “cause marketing” endeavors.
• The definition of “actionable” data will change to include visualization for the non-research stakeholders.  Insights teams will partner with other organizations to deliver results to a wider audience.  The need for storytelling calls for us to leverage research, business intelligence and analytics to create directive results that can be interpreted by a wide variety of audiences.
• Research analysts will increasingly utilize data from non-traditional sources to answer market problems, including customer databases, POS data and web analytics.  This may or may not be part of the “Big Data” evolution, but multiple data sources will increasingly be found in study research.
• Traditional research methods will not die.  Tried and true methods like surveys, ethnographies and the like will be combined in new ways, with yet-to-be-discovered methods to deliver actionable insights for business.  There is no end to the opportunity for change.

I guess we ended up with a “Top 10” list after all! 

2012 will be a transformative year for Market Research.  We at MarketTools are excited about some of the new technologies and methods we are working on, and look forward to finding new tools and partnerships in the months ahead.  Happy New Year!
 

With the year’s end coming, pundits are looking to the social media sphere to come up with their predictions for the newest trends for 2012.  But when it comes to looking for practical direction from the wild and wooly world of social media, I’d advise a little caution. 

Imagine asking a person on the street for directions to a nearby restaurant.  You follow those directions, but you can’t find the restaurant.  You backtrack your route to find the person who gave you the faulty information.  “Hey, mister,” you say.  “Your directions were useless!”  The stranger answers, “But you never asked me if I knew where the restaurant was!”  Social media can seem like it’s populated by people who are more than happy to give you directions even though they have no idea where the restaurant is.

In a previous blog post, I talked about social media in context of divergent and convergent research, the two major market research approaches.  The idea of divergence is to collect and sift through assorted facts, factoids, lies, opinions and hyperbole in order to create hypotheses.  It doesn’t seem unreasonable to assume that the world of social media would help us come up with interesting hypotheses when we are in the divergent phase. 

However, to be accurately predictive, we need to use discernment – the ability to figure out what really matters.  Social media is largely devoid of discernment; it’s millions of mouths speaking at once.  To cut through all that noise, maybe we need a machine like the one Professor Xavier of X-Men used, to scan the minds of everyone in the world to find brainwaves and thoughts that were relevant to him.  We’d have to reach that level of mutative ability to be able to use social media in a directive and proscriptive way.  Until then, we need to be careful about using unfiltered social media feedback and recognize the strengths and weaknesses of the source. 

In my view, the key to making sense out of the volume of social media noise is in the difference between unguided and guided conversations.  I don’t think we have the right tools yet to consistently make sense of unguided conversations, which are the bulk of today’s social media.  I’m much more comfortable using guided conversations – discussions which are lightly directed by a facilitator or a benign dictator – in my divergent research. 

I’m a fan of the online market research communities we run at MarketTools – online communities are a subset of social media where we can create appropriate platforms for focused discussions between our clients and like-minded customers.  Researchers know that they’re interacting with the right people, and they’re able to keep them on topic to elicit the thoughtful feedback they need.

So, as we look forward to identifying trends and crafting strategies to tackle the year ahead, don’t just look to Twitter, Facebook, and the like as your main source of predictive wisdom.  Instead find the right people to talk to and create a place where you can listen thoughtfully, ask questions carefully, and wrap your mind around a narrower, but more comprehensible, set of conversations.

There have been many calls of late for Marketing Research to become more progressive (for example, see the recent blog post from Jeffrey Henning taking off on Coca-Cola’s Stan Sthanunathan’s view that “the industry must change”).  Technology adoption and creating meaningful analysis of social media were hot topics at The Market Research Event (TMRE) held last month, and both the client and supplier sides are experimenting with new trends. 

But while it’s true that the industry needs to move quickly to adopt technologies, blend methodologies, and broaden the reach of research, there is precious little information on how to make these changes.  As a supplier, we at MarketTools have seen that there are a couple of key ingredients in the partnership with clients that make innovative projects worthwhile:

1)    Don’t be afraid to fail.  In one early mobile research experiment, a client was looking for a way to prove the ROI of an event sponsorship.  At the event, a mobile survey was promoted to attendees requesting them to access it via the much-vaunted QR (quick response) code; we also offered a customized short URL.  While the overall project yielded valuable data that could not have been collected without a mobile survey, the tactic of using the QR code to launch the survey was an abysmal failure.  We learned that it’s still very early in the evolution of code reader technology, and almost none of the event attendees had a reader of any kind.  Without the customized short URL to access the survey, we would have had only limited means to collect data.  The flexibility and willingness of the client to experiment with multiple modes of engagement turned a false start into a useful project.

2)    Cast a wide net for executive sponsorship.  When confronted with a large research project, we see many clients getting other departments involved in the project, to spread the resource requirements.  You and your team may need additional backing if you are trying something new, like gamification or visually enhanced respondent engagement surveys.  One client from the research department of a major CPG company was interested in the technical aspects of the visually-driven shopper research experiences we demonstrated.  Because the research department didn’t have a current project that could make use of the technology, we were asked to demonstrate it to the brand team, as a way to answer a question about refrigerated goods promotion – getting the brand executives to back the project instead.  It was an entirely new project type for our research client, and the final study produced evidence the brand team needed to move forward with the promotion.  And our research client gained experience working with new technologies that will benefit future studies.

3)    Understand the statistical requirements.  When the analytics for a project require large sample sizes, multiple choice matrixed questions, or have other complex parameters, then these projects are not good candidates for experimental methods.  There are plenty of cases where clients say, "We want innovative methods, but they need to be validated.”  In these instances, it may not be appropriate to use newer techniques, like mobile and text analytics, that are not set up to have precise parity with pre-existing forms of research.  We find that deeper conversations with our clients to delve into the many different ways that the research results will be used provide guidance on how much experimentation the project can withstand.  For every project whose results must be input into a pre-existing model, there’s a study that’s looking for new input, deeper detail, or variations on a theme to drive a decision or a shift in direction.  These are the ideal projects for experimentation – where the quality and depth of the results are important, yet the business outcomes are not dependent on compliance with entrenched models.

Both suppliers and clients have a role to play in moving the market research industry into its next phase, and this is an unparalleled chance to partner together to discover meaningful directions.  I am looking forward to showcasing more examples in the near future.
 

More and more panel providers are using Facebook as part of their panel management process – we need to be where current and potential panelists hang out in order to recruit and retain great panelists to respond to our customers’ surveys. 

But at ZoomPanel we want to do more than just use Facebook to broadcast announcements or as a channel for one-off surveys, as many panel suppliers do. The ZoomPanel Facebook application provides a complete panel experience within Facebook, allowing respondents to take surveys, answer polls, and check their reward status without leaving the Facebook environment.  Facebook’s interactive capabilities also allow us to interact with panelists and deepen panelist engagement – which in turn helps build longer-term relationships with them and fosters loyalty. 

An example: when ZoomPanel recently launched a new rewards program, we wanted to get real-time feedback from panelists to refine the program during the initial launch phase.  Our Facebook community provided the perfect medium to get real-time feedback from our most vocal user base to refine the program during the initial launch phase. 

Asking panelists to say whatever they want in a public forum about the panel for all to see is risky to say the least – but we decided to explore this avenue to gather constructive feedback to refine the new rewards program into a program panelists wanted.  ZoomPanel directly asked panelists to provide their feedback on the newly-launched incentive program on Facebook. This opened up the opportunity to collect real-time candid comments from panelists who were actively accessing and using the new program.  As soon as the program launched, we instantly found out what was working, and what wasn’t, from the panelist perspective. This quick feedback allowed us to fix bugs and issues in the short term, and make a positive impact on rewards and service for the long term. The instant feedback allowed us to offer rewards better suited to what panelists want, which in turn motivates them to take surveys.

Allowing panelists to speak their mind openly can be a big plus if you manage it correctly or a downfall if you don’t.  ZoomPanel was prepared for the inflow of comments by beefing up and training the support staff on how to add their own comments and avoid vague communication to panelists.  While the positive comments are nice to read and let you know what you’re doing right, the negative comments are very useful and can help guide next steps.  We listened to panelists, made changes and found common ground.  Being proactive, candid, admitting faults and addressing issues right away turned negative attitudes and angry panelists into positive and loyal panelists.  

Using Facebook as an open-book customer service tool allowed ZoomPanel to be reactive and proactive in providing a rewards program suited for our panelists. One of the best things that came out of allowing panelists to speak their mind? ZoomPanelists have become ‘experts’ and now answer each other’s questions, providing a new source of customer support for ZoomPanel!