Market Research – the Life Blood of Social Media

In our earlier post on this topic, we discussed some basics of market research; why it’s important, optional methods, etc.  Today, I want to expand on these basics to give you a deeper understanding of how to do GOOD market research.

  Surveys

Surveys are still a staple of market research because they’re easy to develop and analyze.  Surveys are also relatively cheap, especially with online tools such as SurveyMonkey.  But, done badly, and, like any market research, surveys spell disaster.  Here are some ways to avoid disaster.

Asking the RIGHT questions

It may seem obvious, but asking the RIGHT questions on your survey is really important to get good market research.  That’s why you need to know something about your customers/ prospects before you even begin the process.  And, don’t ASSUME you know what your customers want — ask them.

Often, I’ll use some interviews or a focus group to help me understand what’s important to customers.  I once did a project to a B2B manufacturer to see what customers wanted.  We did a number of interviews and a couple of focus groups first.  These suggested prospective customers were most interested in reliability of the product.  But, the company was only interested in what features customers wanted, so the survey was limited to asking how many wanted each potential feature.  The company then added the features most wanted by customers, despite our report stating customers most wanted reliability, not more bells and whistles.  The company made their product, but it didn’t sell well because it didn’t give customers what they REALLY wanted — reliability.

Coke’s mistake in developing New Coke follows a similar pattern — they assumed sweetness was the determining factor in purchase and set out to create a product matching the exact level of sweetness desired by consumers.  The product was a dismal failure because consumers use other criteria in decision which brand of cola to purchase.

Doing the RIGHT analysis

OK, this is where your statistics professor was right – there are lies, damn lies, and statistics.  In fact, Ross Perot was so good at creating misleading statistics, he almost derailed efforts to approve NAFTA.  So, here’s some advice:

• Correlation does NOT equal causation – the classic example of this is women’s hemlines, which tend to lengthen in economic downturns and shorten in periods of economic plenty.  Of course, hemlines don’t fix economic problems (although it would be nice if that worked because all we’d have to do to get out of this economic slump is wear shorter dresses).  By the same token, economic conditions don’t affect hemlines — it’s not like designers get nervous about the economy and lengthen their designs.  It’s really a function of a 3rd element — which is psychological.  When people get stressed about economic conditions, they tend to be more conservative — with their money and, obviously, with lifestyle elements such as hemlines.  Of course, this has serious implications for the election later this year, but I refuse to get into politics on this blog.

• Scale matters – and Ross Perot as a master at this.  If you blow up the graph big enough, even small changes look impressive.  That’s why your professor stressed statistical significance and why you should also look at practical significance.
• Statistical significance is a measure of the degree to which 2 values are truly different from each other and it’s a complex mathematical function beyond our scope and something taken care of by most statistical analysis tools, such as SPSS.  Let’s just say that it’s impossible to eyeball 2 numbers to tell if they are truly different from each other, so when you graph them,  especially on a really tiny scale, it’s easy to assume a difference that doesn’t exist.  So, with NAFTA, Ross Perot showed graphs blown up so that really small differences in current results and projected results after NAFTA looked very different when, in fact, they were the same — which has proven true after approval of NAFTA.
• Practical significance looks at whether something really leads to different results.  So, something that leads to a 3% increase in customer satisfaction may NOT have a measurable impact on a company’s bottom line — since satisfaction is only 1 element leading to sales of a product.  But, a 3% decrease in price might have a much more substantial impact on sales.  Obviously, the firm would make the changes leading to a price decrease before making changes leading to increased customer satisfaction.