How to Make a Big Impact on Advertising Effectiveness

One of the most famous quotations in marketing is when John Wannamaker said that “half the money I spend on advertising is wasted; the trouble is I don’t know which half”.   Almost a hundred years later, and despite all the analytics and tools, what, if anything, has really changed? 

Below is a picture that might suggest that very little progress has been made. This is a plot of 835 ad executions ranked, from left to right, in terms of each execution’s revenue per dollar effectiveness or contribution.  The 835 executions are unique ads by creative, campaign, ad-length, placement, day-part and platform. These cover a single brand and market. In this plot, we draw lines showing the cumulative impact (as % of total) or contribution and the cumulative spend as we go up the ad count.

Pareto’s Law is Alive and Well

The chart below tells a very common story (we see this in all of our hundreds of engagements). It is clear evidence of Pareto’s Law or better known as the 80-20 rule. Here, for this brand and market, we see that 87% of the media-driven sales is coming from ads representing just 32 percent of the total ad spend or budget.  We also see toward the right end of the chart, that a large amount of the media spend went to ads which generated virtually zero incremental sales.  

Today, many companies are doing media-mix modeling.  However, these models so often are looking at effectiveness at the “aggregate levels” of media channel. It is good to know that Google Ad-words generates more sales per dollar than traditional TV, but looking at media at this level misses so much!

The Devil is in the Details

So, for measuring and understanding marketing effectiveness, “the devil is in the details”. Like shown here, effectiveness needs to be measured and known at the lowest level: by creative, campaign, station-network, ad length and placement, day-parts and platform. While this sounds like too many data points for a model, there are ways to get to this level of detail.  

Once that is done, we do an immense mathematical optimization of these data.  What we normally do is hold total spend constant at current levels, and simply move dollars from the less productive ad-executions to the more effective ones. For our example, a summary of the results are shown in this chart below. In this particular incident, the result showed that this brand could increase its sales by +8 percent with current marketing dollars simply by being more efficient and effective in spending on the ads which drive the most sales!

 A Better Solution for Media-Mix Modeling

This article shows that, despite various tools for doing media mix modeling and marketing attribution, there is still a lot of ineffective advertising being released into the market. With some brands, we suspect that very little has changed since John Wannamaker’s day. The insights coming from aggregate media-mix models do little to guide advertisers in finding which ad executions work best and which ones do not. This all results in a lot of wasteful spending, and with growth harder and harder to come by, this is simply inexcusable.

With media-mix measurement and optimization, “the devil is in the details”. Media measurement and effectiveness need to be understood, measured and optimized at the lowest level of detail: by ad execution.  This means the detail needs to reflect all ad creatives, campaigns, day-parts, ad length and placements and even the stations and networks where they appear. With this level of detail and measurement, there is real & significant opportunity for reversing the effects of Pareto’s Law. This means a stronger link between ad spend and brand growth.

 Michael Wolfe, mjw@bottomlineanalytics.com

Patrick Duncan, patrick@bottomlineanalytics.com