Beyond mere gut feel
Editor’s note: Michael Wolfe is principal of Bottom-Line Analytics, a Marietta, Ga., consulting firm.
Marketing as a function has traditionally followed the “feel-good” paradigm that is rich in intuition and gut feel, one that often has only a vague understanding of what has worked and will work in the marketplace and what the sales and financial returns are on an organization’s marketing efforts and investments. Probably no other statement typifies this paradigm than the famous quote from department store mogul John Wanamaker, who said, “Half my advertising is wasted, I just don’t know which half.”
Today, marketers from small companies to large have the opportunity to adopt a new paradigm. This new paradigm is called accountability marketing and it provides a scientific basis for understanding precisely the effectiveness and returns-on-investment from a company’s marketing efforts. Marketing response modeling (MRM) is a tool that will play a vanguard role in moving organizations toward this new paradigm. The focus of this article is to tell you what is involved in adopting this tool, how it serves the organization’s need to move to accountability marketing and the actual benefits that companies can harvest by adopting this tool and new marketing paradigm.
Operating principle
The feel-good marketing paradigm is one that unfortunately has been the operating principle of many marketing organizations. It basically says that a company can follow a specific marketing strategy and spend its marketing resources not because it has factual knowledge of the results of these actions beforehand but because basic intuition or gut feel says that it is the right thing to do.
Surely, in today’s scarcity economy, with layoffs and radical cost-cutting, companies can no longer afford to follow this path. For one, marketing budgets represent from 20 to 25 percent of a company’s total revenues, often the largest line-item expense on the P&L statement. This places the marketing function into serious peril, as marketing costs become the target of the CFO’s budget ax whenever reality catches up with lofty, indefensible business plans.
The use of marketing response modeling has benefitted many companies. A Marketing Leadership Council survey of companies that have adopted this tool found that, on average, marketing productivity increased 20 to 30 percent! Overall, this means that these companies achieved the same results by spending 20 to 30 percent less, or that they were able to squeeze 20 to 30 percent more out of their present marketing spending by using the guidance that these models provide.
Five steps
There are five essential steps required for a complete market response modeling project. These steps involve the following:
1) A “wheel of fortune” exercise to establish data requirements and a plan for collecting essential input information for the modeling exercise.
2) Building and validating the models. This is a technical phase, but models must demonstrate proven validity for the task at hand. Validating models often requires holding out certain periods of data and testing how well the models forecast over those periods.
3) A model diagnostics phase. Once valid models are constructed, this exercise amounts to a diagnosis of the effectiveness of recent and past marketing efforts and programs. Included here is a linkage of model output to financial data to determine current marketing returns-on-investment.
4) The third phase is marketing optimization. Here, you would link current marketing program spend with the marketing driver volume contributions from the models. In doing so, you can derive efficiency ratios to determine how much volume per dollar spent each marketing element is presently delivering. Then you marry modeling-based sensitivities to marketing spend and determine, with a given marketing budget, the ideal or best-practice allocation of marketing funds to drive volume, revenue or profitability.
5) The final phase of the process is referred to as strategy simulations. Here, you marry forward marketing plans with model sensitivities and actually simulate or forecast the impact of any given plan. Once that is done, you will also need to link the results with financial and planned spending data in order to arrive at an assessment of the net incremental profitability of a current plan.
Step 1: The wheel of fortune exercise and determining model data requirements
First of all, it must be understood that going through a market-modeling exercise requires due diligence in specifying and collecting the required information or data. Most often, these data will be a series of metrics over time, which describe both the output (sales or revenues) and the key marketing or other drivers of business performance. Frequently, this begins with a white-board type of exercise with company internal information and systems experts. With your desired result or output at the center of the hub of the wheel, you then construct the spokes to include a comprehensive list of marketing and other drivers of the business.
An example of this is illustrated in Figure 1. This example displays what would be the key business drivers for a soft drink brand such as Coca-Cola and is also similar to what you would see with other consumer products businesses. It is important to note that for other industries, the drivers might be different. For example, for an insurance business, agent hires and agent compensation schemes might be an important driver. Likewise, it is important to note that not all business drivers are necessarily marketing related or within the control of the company. Some of these might be external factors such as the economy or weather. Or, perhaps like the airline industry, the 9/11 terrorist attacks were an event that had a definite and measurable impact on that business. These types of factors also must be included, along with their metrics.
The key to this exercise is that, once the wheel has been constructed, you must have explicit metrics for each element. These metrics should be measures of the level and intensity of each activity as they are executed in the market. This could be expenditures or it could be some other metric, such as advertising gross rating points. Once the data has been specified, all of these need to be integrated into a single database covering the entire historical period in which the model is to be constructed.
Step 2: Building the models
Market response modeling is a data-intensive and data-driven exercise. It is critical to get it right up front. As the saying goes, “Garbage in, in garbage out.” While this step requires a resource with the due technical expertise to construct the models, it is important that companies require that the model-builder provide them with sufficient validation in order to underscore the model’s validity and integrity. What is usually required here is what we call a holdout validation. In other words, about 10-20 percent of the data are to be withheld from the modeling process and validation entails a comparison of how well the model predicts sales over the unknown periods.
Step 3: Market response model diagnostics - What’s working well and not so well?
Once models are completed and validated, then we move into the diagnostics phase. This phase focuses on current and past marketing initiatives and tells you what is working and what isn’t. By identifying the not-working components, marketers will gain the insight to mine marketing spending of considerable savings.
One of the key outputs from this analysis is a sales decomposition analysis such as the one shown in Figure 2. A sales decomposition analysis conveys a simple message. In a pie chart as shown here, it tells you the proportion of your sales that each marketing element, such as promotion and advertising, has delivered to your total business. Note that marketing does not represent 100 percent of sales. The difference between total and marketing-driven sales represents a term called base volume. Base volume is essentially the estimated level of sales that would exist in the absence of these marketing activities, while the total marketing contributions represented here shows you the total risk to your business if your company ceases its marketing efforts.
A second type of marketing diagnostic is what is called marketing variance analysis, which essentially tells you what is driving your current business performance and trends and by how much, or simply put, how much of your current performance is due to advertising, pricing, etc. A sample of this is represented in Figure 3. In Figure 3, you can see the degree to which retail price and advertising are driving current performance for this brand. This analysis is the definitive scorecard regarding what’s been working and not working in the marketplace and how effective the company has been in executing the present plan.
Another component of the diagnostics phase involves marrying model output information from the sales decomposition with marketing expenditures and brand profit margins in order to derive a complete marketing return-on-investment analysis. An example of this type of analysis is depicted in Figure 4, which shows a comparison of the relative costs and financial returns for four major buckets of marketing spending for various forms of advertising and promotion.
Step 4: Marketing spend optimizations - Where best to put those dollars?
The next stage requires merging model response information with actual marketing expenditures for each marketing driver represented in the model. This also requires accounting information pertaining to spending for each key marketing program.
As shown in Figure 5, the concept is simple and straightforward. Visually here, you would compare model-driven output volume for each marketing element with its corresponding expenditures. Here you can compare advertising and trade and consumer promotion spending with their relative volume contributions derived from the models. As you can see in the example, advertising was truly driving more sales than its proportion of the marketing budget would suggest. Based on this, you then extend this analysis by coming up with an optimization of the marketing budget, which provides you with the mix of spending that drives the most volume or profit.
The optimization phase clearly is one of the most important because it provides clear direction on how best to spend a company’s marketing budget. Not only this, but it provides a view as to the degree of inefficiency in the present marketing budget and can quantify the benefit of moving toward a more ideal or optimal marketing spending plan. Experience indicates, as a rule of thumb, that somewhere between one-third and one-half of all marketing expenditures can be classified as non-productive or highly inefficient. Clearly, with companies spending from 20 to 25 percent of their total revenues on marketing, this represents a substantial saving opportunity as well as a tremendous overhead.
As a final exercise here, you can compare what sales, revenue or profit opportunity exists by moving from the present marketing spending scheme to one that is more ideal or optimal. An example is illustrated in Figure 6. For this business unit, there is an opportunity to grow by an additional 10.6 percent if management can redeploy current marketing resources towards the more efficient or ideal mix suggested by the model. For almost any business, an additional 10 percent of bottom-line growth surely represents a substantial sum - one that may very well mean the difference between zero and substantial bonus payouts at the end of the year!
Step 5: Forward strategy simulations - forging a go-to-market plan for maximum payout
The final stage of the process involves exploiting the market response models in order to take a forward look at the likely consequences of future business plans. Figure 7 illustrates an example. In this instance, a five-point strategy is laid out involving launching a new soft drink package, a price rollback, expanded advertising spending and the expansion of in-store coolers. From the model, you can then document the expected growth contributions from each action, as shown.
At this final phase of the process, you have now successfully outlined a go-to-market business plan. Not only have you identified the upside volume potential of the plan, you have also isolated the contributions from each component. Such a simulation provides a future view of the results of planned actions and gives a specific rationale for that plan - a justification for upper management, if you will. Once this step is completed, then only one more exercise is recommended. That involves sharpening the pencils and doing the due diligence of turning this into a full-blown financial justification.
Moving forward with the accountability marketing paradigm
As shown in this article, market response modeling tools can provide every company with the capability to move towards a new marketing paradigm. Tools such as market response modeling have shown substantial and verified benefits in those select companies that have chosen to take the leading edge.
For example, at the Coca-Cola Company, a number of instances can be cited where substantial returns were uncovered through the application of market response models. In one Coca-Cola division, the company was facing a dilemma with a money-losing returnable plastic package. In fact, the company was losing nearly $7 million on this package. After completing a market response model exercise, the challenge was to come up with a simulation that allowed the company to discontinue this package, perhaps launch a new package and yet not damage the total volume for the company brands. To make matters worse, at the time of this exercise, the company’s total volume was declining by 16 percent.
As a result of our modeling, model simulation was used to recommend that the company discontinue the unprofitable package and launch a new plastic one-way pack. By doing so, it was estimated that total volume would increase by about 11 percent. One year later, this division actually implemented these recommendations as presented. The unprofitable pack was discontinued and the new pack introduced. As a consequence of this action total volume was not up 11 percent, but rather 13 percent! In consideration of the $7 million in savings and the additional profit generated through growth, a total of about $14 million dropped to this division’s bottom line!
A few years prior to the above, a market response model project was conducted in a tropical country which, at the time, was showing the weakest Coca-Cola case sales performance among 23 peer countries within its greater business group. The challenge at the time was to come up with plans and recommendations that would hopefully reverse the fortunes of this country. As model simulation exercises were completed, an opportunity was discovered to grow the business substantially by placing new coolers in some large and selected cities and channels (actually for this developing country, the largest channels were mom-and-pop stores and street kiosks).
The second part of the recommendation focused on expanding availability or distribution of Coca-Cola company secondary brands (such as Fanta, Sprite and some local brands the company owned). After the recommendations were made, the performance of this country’s case sales was traced. One year later, the country had improved from 23rd to ninth place in case sales performance within the business unit. After 18 months, its performance had improved even more: it was in first place among the 23 countries. Not only did the country’s business performance improve substantially over this time, but the sales improvement was directly linked to the execution of the recommended two-point strategy! Overall, the benefits of this action were substantial. Given the swing factor in case sales, the total improvement in the country’s profitability actually increased by about $9 million.
Might seem daunting
While the task of building predictive models might seem daunting, the only requirement is that companies have end-customer sales data and calendars of when and how much was spent on various marketing initiatives. This could happen in either consumer or business-to-business markets, for large companies or small.
The benefits to companies who travel down this road are many. These include identifying your unproductive marketing programs, with savings potential in the millions. In addition, companies will be able to identify business opportunities and quantify both the sales and profitability of those current business strategies. What’s more, companies will get fact-based guidance on how to best allocate marketing-mix spending across sundry marketing programs that are likely to yield the largest net results.
In sum, moving from the feel-good marketing paradigm to the accountability marketing paradigm might require changing one’s mindset about how marketing operates within your organization. Nevertheless, the benefits and necessity of moving towards the new paradigm are clear and compelling for large companies and small.