Yes, the purchase funnel as a linear model in a non-linear world, is a big problem. My own shopper research done in partnership with InsightsNow on behalf of AOL proves that a high percent of brand consideration now forms as people are doing their pre-shopping research or right at the point of purchase. McKinsey found the same thing.
However, the BIGGEST problem with the marketing funnel is that it leads to the media funnel. Because we think awareness comes first, then consideration then finally purchase, we move from big to small, from reach to targeted advertising, from attitudes to behavior.
I propose that marketers flip the media funnel. We need to start with desired behavior (measured as “conversions”) and work back through digital behaviors and ad impression serving that are most likely to lead to these conversions. It is media planning in the era of big data and data science and it works in the other direction.
In the attached diagram consider media placement in terms of A, B, and C levels. Current Media planning practice typically moves from C to B to A. Now let’s think about planning media by flipping the sequence.
The average elasticity to ad spend is in the .1-.15 range (i.e. double your spending and get a 10-15% increase in sales) according the the Wharton future of advertising project. Consider that as a grand mean. I hypothesize that advertising elasticity sub-means will be highest for A, second highest for B then comes C. If so, we should be spending as much as possible on A opportunities, then move to B, then finally to C to round out the reach we hope for. This should result in substantial differences in how marketers spend their money and what research is needed to add value.
The key questions are:
1) What is the relative impact of impressions at each level of targeting?
2) Assuming that A level targeting produces the highest conversion rates, how can the marketer differentiate when someone is available for an A level of targeting?
3) How much scale can you achieve with A alone, or with A+B targeting?
“A level” targeting needs to be better understood. How can you spot if someone is in an A moment? This is a statistical question best addressed via predictive analytics but variables I would consider, based on theory are:
- Indicators that someone is in a pre-shopping research phase or about to transact
- Indicators that someone neither loves nor hates your brand…that they are in the middle of the U-shaped curve of preferences (i.e. probability of buying your brand) and so they are most persuadable.
In digital life, there are lots of signs as people pull content that is important to their shopping mission. In shopper work I have done, I found that over half of shoppers turn to digital tools for most products and services. Even grocery products experience a large percent of shoppers using digital. They search, visit websites, visit retailer sites and explore their potential purchases, they visit store locator pages and comparison sites, and of course, they go to physical stores. Use predictive analytics to identify these moments. You can imagine analyzing a database where we know who shopped when and, working back through the clickstreams, using logistic regression with a bunch of digital breadcrumb variables to determine the strength of signalling.
B moments are also very interesting, as there is likely to be a conversion lift vs. C moments. Also, B moments will be needed in the media plan to achieve the reach that marketers want. B moments occur when a user is browsing based on their interests and lifestyles without necessarily having immediate shopping purpose. In the AOL work, we found that a quarter of people regularly engage in this behavior.
The final piece of advice I have is that marketers need to expand their A and B opportunities by creating brand audiences based on brand liking and compelling content. These first party data and the A and B moments they create will be worth their weight in gold. Some brands have tens of millions of pageviews per month to their website because they have created a community where people can find relevant content. A retailer can have hundreds of millions of pageviews each month. It is likely that Facebook fans are mostly in the middle of the curve so updates directed to them is good.
The targeted nature of A and B moments make it likely that C moments will be needed to top off the reach the marketer desires. However, to my knowledge no marketer or agency has tried to measure the reach of A and B moments and so we do not know how much of C moment money is being wasted. To improve advertising productivity, this must be known.