What Brand Attributes Make a Difference

When David Ogilvy says, "Any damn fool can put on a deal, but it takes genius, faith and perseverance to create a brand," many of us nod knowingly. You spend an overwhelming amount of time (and in some cases money) defining, building, and executing on a brand that will drive your business forward. That's why you run brand workshops, that's why you create style guides, and that's why you obsess about refining a list of brand attributes until it perfectly captures who you want to be and what emotions you want to elicit from your customers. That master list of brand attributes will impact every touchpoint between your business and your customers, so it's worth getting right.

But how do you know which brand attributes matter the most? How can you be sure that you're emphasizing the brand attributes that will most effectively lead to business results? Brand is a slippery, amorphous thing, like a fish that flops around while you try to weigh it. But because brand attributes drive so many decisions, from the tone of all your messaging to the design of your website, it's critical to find ways to measure performance beyond that blunt instrument called a customer satisfaction survey.

An emerging technique for prioritizing brand attributes involves correlating each attribute to increasing share of wallet with customers. The goal is to identify which brand attributes are most closely linked to customer loyalty by looking at how much customers spend with your business out of all the money they spend in your industry. By examining which brand attributes are working most successfully with the most loyal customers, you get compelling evidence that these are the brand attributes that are most important to your business.

Here's an overview of how this process works:

  1. Survey customers on current brand perceptions. As part of a customer survey, ask customers to rate on a scale how much they agree or disagree with various statements about your company or website, where each statement focuses on a single brand attribute. (If you don't yet have a list of brand attributes, a brainstorming workshop with key stakeholders is called for.) Examples might include "Company X is fast," "Using Company X is easy," and "Company X has great quality." This is also a good opportunity to test off-brand attributes as negative statements, such as "Company X is unreliable" and "Company X is only for experts." Test as many positive and negative primary and secondary brand attributes as possible, and make sure to randomize the order of the attributes.
  2.  Survey customers on share of wallet. Because you want to measure how much customers spend with you relative to their total spend, you need to ask this in a survey rather than rely on customer financial data you have. While customers aren't necessarily accurate in estimating how much they spend, they are pretty good at estimating the relative amounts they spend with different companies. Have them enter total spend in your industry and then total spend with you. For someone who spends $750 with you and $1,000 total, you have a 75 percent share of wallet.
  3. Conduct statistical analysis to find correlations. Grab a statistical analyst for this step, so he or she can run the appropriate analyses on the data you've gathered. Segment your survey respondents based on their share of wallet (e.g., one segment with less than 25 percent share of wallet, one with 25-50 percent, and so on)-- we usually do quartiles or deciles for the segments. Then look across these segments to find those positive brand attributes where strong agreement correlates with an increase in share of wallet and  those negative brand attributes where strong disagreement correlates with an increase in share of wallet. For the statistically minded, this is done using linear regression. You'll see that some brand attributes bubble up more than others to correlate to share of wallet. Which brand statements are very loyal customers most likely to agree with? Those attributes are important and may be driving business results. Meanwhile, some off-brand attributes will be more important negative correlations. A strong negative correlation on unreliability means that reliability could be a key brand attribute.

What emerges from this process is a list of brand attributes that is prioritized based on measurable business results. What was formerly the touchy-feely realm of branding now has a little more science injected into it. As any statistician will tell you, correlation is not necessary causation, and this data doesn't prove with certainty that certain brand perceptions cause higher loyalty. But this data is a strong indicator that certain brand attributes matter more than others, and thus can help you prioritize and focus your brand.

All sorts of decisions become easier after prioritizing brand attributes. Rather than juggling a long list of diverse brand attributes when writing email copy or choosing a color palette for your new website, you can now focus your team on the few primary attributes that are most important. In addition, testing new concepts becomes easier, because you can have customers evaluate a concept against these prioritized brand statements, just as they did in the survey.

Brand attributes are the winds behind the sails of many decisions that you make. Isn't it about time you measure which gusts of wind are most successfully moving you forward?

Steve Mulder and Ziv Yaar are colleagues at Molecular, an internet consulting firm in Boston, where Steve is principal user experience consultant and Ziv is vice president of internet strategy. They are the authors of "The User Is Always Right: A Practical Guide to Creating and Using Personas for the Web."

 

Comments

sid Mohasseb
sid Mohasseb May 5, 2009 at 11:02 PM

Dead on –

The key is not to guess on the brand attributes but to discover them as well as the relative significance and level of positive negative sentiments. I would like to invite you to check out www.wisewindow.com – I would love to hear your thoughts on auto discovery using the social media. Sid

ENumerys Global
ENumerys Global October 27, 2008 at 3:34 PM

Is the assumption that consumers are good at providing some form of proxy for SOW or is there actual empirical evidence? Maybe higher ticket items, reported may even come close to actual SOW, but for FMCG products, I would expect consumer reported estimates on spending (SOW less likely) to rarely come close to actual. For instance, for batteries, if you ask me what percent I spend on Duracell vs. Energizer vs. Rayovac vs. Eveready, I will probably be more than a little perplexed. So why not just use brand-tracking data and correlate (or even run a multivariate model) that to actual sales? You don't have to worry about reported vs. actual consumer behavior and you will end up with a much better sample size. Agreed aggregate data may not give as much insight as individual level, but on the flip side you have more consistent estimates coming from a larger sample size vs. surveying a few thousand consumers as a proxy for millions.

janice lane
janice lane April 16, 2008 at 9:42 PM

brand correlation is importannt

Laura Dascalescu
Laura Dascalescu April 3, 2008 at 4:46 PM

I think this is a very interesting article, and I am looking forward to reading more of them.