Byron Sharp’s How Brands Grow is the only marketing text I refer to on a regular basis. It’s well written and pulls no punches. His conclusions are based on single source datasets tracking attitudes and purchasing over time, giving them rigor and power.
The evidence he presents on the unstable nature of attitudes is particularly useful (p102). Brand imagery statements we all know and love (e.g. “Is a brand a trust” “Offers VFM” “Tastes nice” etc) often show broadly stable agreement over time. Sharp’s insight is that when you ask the same people the same question over time their answers are often different. The repeat rate for attitudes is around 50%. So for example if 45% agree brand x tastes nice in Q1, and 45% agree in Q2, evidence from single source datasets show that half the people “voting” for brand x have changed their minds since being asked previously.
Sharp talks about the “fickle, probabilistic nature of beliefs” which are relative, not absolute. People tend to buy from a repertoire of brands, and our beliefs are dependent both on our power of recall and the situation. As he puts it “I might choose Avis 60% of the time but that doesn’t mean that sometimes I won’t choose Hertz even two or three times in a row…” There’s a range of other published evidence showing similar findings (e.g. here and here).
As a market researcher you miss this meta-level reflection. Ad hoc work takes us from project to project, where we burrow into hypotheses then move on. Tracking tends to use fresh samples for each wave. Client work is confidential.
As an industry we suffer from an acknowledged, accepted, scientific body of knowledge upon which to base much of our work. Ask two people how advertising works and likely you’ll get two different answers. Econometrics or agency ad effectiveness models re-create a version of reality, not reality itself. From time to time it’s hard not to feel like William Gull as described by Adam Curtis “…a mad figure seeking to impose narrow order on a fluid world that he can only partially understand.”
I take two implications from the evidence on attitude repeat rates.
Firstly, to ground our research findings within an appropriate context. Peer influence on choice (e.g. what others are doing) and convenience (e.g. location, ease of buying) are good places to start.
Secondly, it’s a reminder not be over reliant on brand imagery batteries when diagnosing brand health. Awareness and purchase behaviour are as worthy diagnostics.