What Cross Tabs is teaching me about market research
At the start of the year, I launched a project called Cross Tabs, a podcast (and newsletter) about how polls are designed and deployed, what they tell us and what they don't, and how to make sense of them in yet another insane election year.
I set out to understand how polls work; candidly, I thought of polls as a kind of cousin to market and marketing research — related, but not the same.
What I've learned, however, has made me rethink how we approach market research in our practice; and that, in turn, is influencing how I think about brand and go-to-market strategy for the clients we work with.
Here are a few things I'm thinking about now as a result of talking to a host of experts in this space across our first seven episodes.
More data doesn't necessarily help you get a more accurate picture of your audience or customer. David Radloff talked about this in an early episode. He works on turnout models for campaigns at Clarity Campaign Labs; despite the availability of loads of data sets that can be combined to get ever more granular pictures of a given voter, only a few factors turn out to be determinative, and the additional data might be window dressing, or it might be a distraction from the useful truth.
Local knowledge is critical for building models of behavior. Amelia Showalter at Pantheon Analytics talked about this in our interview in the context of designing electoral maps. Without understanding the actual lay of the land, and how people navigate it and build communities within it, you can't effectively draw boundaries around them. There are a lot of theoretical "communities" these days, but without getting involved with them, visiting them, going and seeing them for ourselves, we are likely to get our definitions and descriptions of those communities all wrong.
Mixed methods are better. You can't learn everything you need to know with a single method. It doesn't matter how comprehensive your discussion guide is, how long your survey is, how representative your sample is. You have to use multiple methods to really understand something. Do qualitative research before you do quantitative. Use digital ethnographies and in-person observation in addition to Zoom interviews. Conduct attitudes & usage studies or full-on segmentations; conduct concept tests; ask people questions more than once. My conversation with Natalie Lupiani and Shannon-Janean Currie at BSG really drove home the importance of mixing it up.
Quality data is getting harder to come by. In my conversation with Paul Westcott about L2, Inc. 's work on the voter file, and in an upcoming conversation with Rich Ratcliff at OpinionRoute about online, opt-in (aka "convenience") sample, I've been stunned by the lack of concern that a lot of market researchers (until recently, myself included) had about the quality of the respondent pool we get from sample providers. While — now — it's a lot easier to vet qualitative respondents, a variety of factors have combined to make high-quality sample harder to get. I'll write more about this soon, and look out for my upcoming conversation with Rich for more really mind-blowing detail about programmatic sample.
Quality data is more important than ever. Before these conversations, I might have consoled myself that even directionally useful data is better than no data at all. But I'm becoming less convinced of that as I learn more. We can't make decisions based on fraudulent or bogus responses. Businesses are operating in a time when the cost of money is comparatively high thanks to high interest rates; consumers are operating in a time when a lot of things are more expensive thanks to high interest rates and high consumer prices on a variety of goods... and of course, thanks to the insanely high cost of housing, healthcare, and education. I saw a survey the other day of 18-29-year-olds that showed half of the respondents said they were homeowners. There is simply no way that is correct; the average age of a first-time home buyer is about 33. If you're about to make decisions for your brand or your products based on a sample this skewed (or this bogus), you're about to make some very costly mistakes. One way to check your data is to use other sources. Your survey is not a single source of truth; government data, academic research and other publicly available sources can help you gut check your research (and maybe also help you save yourself time on a questionnaire when you can find better quality information via other sources).
We're due for an update of our theories. I see this in the Ritson v. Sharp debates on this platform amongst my adland friends; I see it in conversations I've been having with political scientists and political psychologists like upcoming Cross Tabs guests David C. Wilson Ray Block Jr. and Dr. Camille Burge at Villanova. We have been coasting for a while on theories that are sometimes more like folklore. One idea whose time in the barrel has come, in my opinion, is that of the "high involvement" purchase. Any textbook will tell you that "complex" and high-cost decisions require providing lots of information and service during a period of "extended problem solving", including reassurance after purchase. But it's increasingly our finding that what is a "high involvement" purchase for one person could be a "low involvement" purchase for someone else; and that they are highly contextually dependent. We make lots of expensive decisions without any visibility into the details of what we're buying, including the price — consider the healthcare sector, as an example. In other contexts, ideas like measuring affinity for a brand, "racial resentment" among voters, or "narcissistic traits" among college students all need to be much more deeply theorized and studied before we continue to use traditional tools for measuring these phenomena.
I'm obsessed with salience. At Hall & Partners, they used to have a model known by its factors: Promotion-Persuasion-Involvement-Salience. In this model, promotion meant sales and other offers; persuasion meant rational benefits ("gets your whites whiter"); involvement meant an emotional connection or shared values between consumer and brand; and salience was often measured by this statement, "It's a brand that's on it's way up and really doing something." But as I've talked to people like Mike Podhorzer and others, I've become increasingly convinced that salience is better defined as "the issue/characteristic/feature that compels me to notice, consider, care, act". That characteristic may operate functionally/rationally, it may operate emotionally or socially, or it may leverage in-group/out-group dynamics or play on your fears more than your hopes. But it is the feeling of something being important enough to you that you are compelled to think... and then act. In my view, this is the most important response any execution of a campaign strategy can summon. If you're a challenger brand, a new entrant, or a big player trying to steal a little share on the margins, the hardest part is to break through to the consumer on auto-pilot, satisficing her way through a purchase, and get her to notice you and think about you. But here's the thing — what's salient to you, the founder, the CMO or the brand manager is, in my experience, almost never what's salient to your prospects and customers. Figuring out what is, and how to catch that wave is everything.
I'll be writing and thinking a lot more about this. I think salience is the key to habit/loyalty, it's the key to preference, it's the key to fandom, it's the key to rejection or opting out of the category altogether. Salience is your most important brand and messaging metric — and it seems salience is the key to getting out the vote, too.
I'd love for you to listen and subscribe to Cross Tabs. You can find it in any podcast feed. And you can subscribe to the weekly newsletter at www.crosstabspodcast.com.