Modeling the Return on Media and Message Investment

Seabuscuit and sales

In August of 1936 auto magnate Charles Howard paid $8,000 for a horse that at three years old had already run 47 times and won less than 20% of the time. A scant four years later, in arguably the match race of the century the colt had beaten triple-crown winner War Admiral, became a beloved legend garnering more headlines that year than the then three term President Roosevelt and returned more than 50 times his selling price in winnings, not counting future stud fees. In advertising, at an average 5% advertising to sales ratio, we settle for a 20 times multiple. But like horse racing, it is often difficult to predict the outcome of a campaign, though we are getting a lot better at it.

The combination of single source research, mathematical modeling and advertisers’ willingness to spend money for answers has advanced the art and science of predicting sales outcomes for different advertising scenarios, for both media and message strategies. The model described in this paper accounts for not only for sales driven by paid advertising, but also media inspired brand equity sales.

Betting on the front half of the action

One of the problems with current modeling techniques is that they deal with only half the action. Once they have identified the extent to which brand equity is responsible for a sale, they ignore probing it for any more detail. Unfortunately, for most major brands, brand equity accounts for well over 50% of sales. Nonetheless, the statistics modelers deal with are confined to advertising driven sales, which by their very nature are immediate, not total sales (and usually less than half the total). This is like looking at Seabiscuit’s record before he was sold to Howard and closing out his career as a sad three year old. It ignores what was hidden, the equine’s equity. The colt was a descendent of Man ‘o War, was raced to that point to pace stable-mates and not really given much attention by previous trainers. But Howard’s trainer looked at the whole package and saw the potential for something quite different. Likewise, in order to properly evaluate the potential contribution of a medium or a message to sales, one has to appraise its return in terms of both immediate sales (the front half) and also what it contributes to brand equity and therefore ultimately to future sales (the back half). This article describes a technique that analyzes, models and refines media and message strategy to return substantial incremental “total” sales (including brand equity), without spending one extra penny.

Fragile inputs/reasonable outputs

With the understanding that the entire enterprise of statistics and probability is really a very sophisticated fiction we impose on the real world in an attempt to describe tings in a quantifiable way, we looked at the current state of the art in identifying both media and message return on investment for total sales. An example of statistical fiction and an imposition we accept is the currency we use to negotiate television time, cost per thousand (CPM). High level network sales people, agency negotiators and client media directors argue over nickels and dimes, sometimes pennies, in an attempt to negotiate the pricing on network packages worth sometimes hundreds of millions of dollars. Because the ratings system we have in place is the very best we can do for the time being (some 50 years now) in gauging the value of various programs, we accept the CPM convention. The truth is that the statistical error swing for a primetime CPM (while we are negotiating for pennies) can be as much as $4 and that does not even describe how far the estimate is from reality. It merely illustrates that if the package were re-sampled 100 more times, the answer would likely come back within a $4 range (+/-$2). So services that track and model a client’s sales and marketing initiatives on a weekly basis are doing the best they can and are now able to attribute a sales to a specific media or message campaign, using accepted statistical techniques.

The equity equation

So current modeling techniques assume that brand equity plays little to no role in identifying a favored campaign medium or message. Models are generally all about immediate sales, provoked by very recent advertising (recency). But to assess the full value of a medium or a message marketers should really account for their influence on brand equity as well, which in turn inspires future sales. Brand equity is comprised of many marketing inputs like paid media advertising, promotion, public relations, packaging, in-store displays, pricing, method of distribution and word of mouth among others. Modelers can tell us what proportion of sales are due to brand equity stimulus as opposed to recent advertising and it is usually more than half. Then based on a combination of objective client research and information, as well as subjective experience, and with the understanding that advertising can be a strong force in the equity mix, we can now deduce a reasonable range within which media influences brand equity. It is then fairly easy to attribute a value to individual media types. So the modeler assigns a portion of immediate sales to individual paid media and this new model ascribes a portion of equity driven sales to the medium as well. The two can be added together yielding a complete picture of the medium’s contribution to sales and therefore its full value. This procedure enables us to better gauge the full value of a medium’s ROI.

The key: a cost per sale

After arriving at a medium’s total sales (immediate and equity driven), it is just a matter of aligning those sales with the number of target points (trps). By dividing trps by sales, one can derive the number of target points it takes to make a sale.

From the client, it can be a simple matter to get the appropriate cost per points for media in the campaign. Then, by multiplying the target points per sale by the cost per points, you can arrive at a cost per sale. This figure, the key to ROI, can be compared across media or if one knows the media mix for a particular copy campaign, by prorating these figures across media you can get a cost per sale by different copy strategies.

Using cost per Sale for ROI scenarios

These total cost per sale figures for a medium or a message can be used to run payout scenarios. For example, if one copy strategy is 10% more effective in generating sales than another and you shift $20 million from the lesser campaign to the more effective one, the model indicates that within a short time, $2 million in incremental sales will be derived. Or if one medium is 25% more effective in generating sales that another and you wish to switch ½ the weight out of the more expensive medium (say $20 million), the model indicates an incremental sales gain of $5 million.

Of course the model also has to correct for the overlapping influences of the various changes being made (like the combined effects of media and copy mix), so that they are not purely additive. It also must correct for the diminishing returns of employing the same strategy over the long haul. Nonetheless, even with these corrections, the incremental sales gained in the short and longer term are very substantial. One advertiser increased sales by more than $100 million (example attached).

Other uses

This model also enables us to examine other phenomena like the effects of hiatuses, delayed responses to advertising, comparisons of a medium’s cost per sale to the value of a sale, the difference between a medium’s effectiveness in the short vs. the long run, and media weight and shifts. The only input necessary is a brand history from the client, the media plans (with costs), data from a company that has tracked and modeled sales and media and messaging and the proprietary equity and ROI models described here. It is also possible to deploy the same models on other marketing initiatives, which draws us all even closer to breaking the code on the DNA of investment return in the marketing of all products and services.

$100 Million Example

<back to top>
<back to Essays>

© Media Directors Ink : September 2003