All CMOs that I have spoken with agree that a big issue they grapple with is how to balance their investment in brand vs. performance marketing.  I’d like to take a run at this question a different way by starting with the observation that short-term effects and long-term effects of advertising mostly come from two distinct populations of consumers. 

If so, this CMO-led debate should be about targeting as much as what creative to run, should we use TV, digital display, etc.

Consider a long purchase cycle category…new vehicles.  At any point in time, actives (currently shopping for a new vehicle) are, say, 5% of the population of drivers, with the remaining 95% being dormant (not currently shopping).  Ignoring some impulse purchasing, all of your short-term sales lift will come from the 5% of the population who are active.  Dormants are probably the main driver of any measurable long-term effects as portions of them cycle into becoming active over the next 12 months. Research and analytics folks: when you have a hard time reconciling campaign results in terms of sales modeling vs. survey-based brand awareness or favorability, keep in mind you are actually comparing two distinct effects from different populations.

In CPG, things are a bit muddier for long term effects as they can also come from repeat purchases off of a short term incremental sale. But the short term effects pattern is still clear (and probably universal); advertising to actives (those about the buy the category) will, in fact, produce 2-3 times the short term ROAS. (The Persuadables).

This active/dormant framework suggests, if you want to drive more short-term sales, use targeting to deploy a greater percentage of ad impressions to actives (or what might be called “intenders” by AdTech and media partners), especially if you target the hyper-responsive Movable Middle (white paper here.).  This is irrespective of creative, so go ahead and use your best brand building creative to drive short term sales! Reach based strategies deliver advertising to a broad population of category buyers…so you are sacrificing a predictable sales bump today in the hope that it will generate improved brand favorability and that will lead to somewhat bigger long-term effects; the hope is that this nets out in your favor.  

So now we see you cannot solve for maximizing both short term and long term effects (unlike what a head of media research at a large CPG once incredibly told me!) But, what is the right balance between ad impressions served to actives vs. dormants?  And what tactics and messages work best against actives vs. dormants?

We don’t know!  Marketers guess at what is the right blend.

What we’re missing in today’s brand research and modeling protocols is longitudinal single-source tracking of consumers’ brand attitudes, exposure to advertising, and long-term conversion behaviors. We need to study how advertising increases brand favorability and in turn, the rate at which that translates into an annuity of future sales. 

This gap will be addressed by The MMA’s research program called “brand as performance” (BaP). For each test case, we will be identifying a large backbone of persistent IDs whose ad exposures and conversions are trackable over an extended period of time and who are also on a survey panel so we can measure conversions by brand favorability, within exposed vs. unexposed consumer IDs.  We leverage MTA and test vs. control protocols to analyze the data.

If interested in the MMA’s BaP research program, please contact me joel@rubinsonpartners.com

One final observation for those in research and analytics…I’m not sure that adbank (the recollection of prior weeks of advertising, discounted at some geometric rate over time, added to current week’s advertising) is the right way to look at long term effects of advertising.  I would suggest that advertising today can improve brand favorability today. It is the durable memory of brand favorability that feeds into the consumer mind which is a predictive engine for how to best satisfy the needs they are shopping for. If there is no change in favorability, there will be no long term effect (or you are not measuring favorability correctly as I often see with exotic brand equity metrics). 

Note: In addition to blog posts, I have published a number of peer-reviewed papers and white papers that I have made freely available in a repository here.