10 Facts About Marketing Budgeting

What to take into account when determining the budget for advertising or other marketing activities to avoid mistakes.

By Sönke Albers

John Wanamaker, an American retailer and marketing pioneer, once said, “Half the money I spend on advertising is wasted; the trouble is I don’t know which half.” This shows in a very pointed way the dilemma of marketing budgeting. If I do not invest enough in advertising, then I cannot expect to achieve the revenue I had hoped for. On the other side  if I invest too much money, it will have no impact, which in turn basically means to throwing it out the window. So what should you look for when setting your budget for advertising or other marketing activities to avoid making mistakes?? The next ten points address precisely this question.

1.you must Know something About the Effectiveness

Setting marketing budgets is in many ways a highly emotional process that hinges on the negotiating skills of those responsible for product management. This is shown in (3) for Bayer Pharmaceuticals. However, it is important to know as a product manager for the negotiated budget that you also need to deliver the promised results. This is why it is advisable for everyone involved to find out how effective certain marketing activities can be, be they advertising, sales force, or distribution. In the case of advertising, for example, it is not enough to determine how many people in a target group you want to reach and how much it will cost. This always raises the question whether you could have achieved better results with more or less advertising expenditure. An answer to this question can then only be found in the empirical knowledge from your own data analyses or from meta-analyses of studies which have already been published on the effectiveness of marketing activities.

2. Can you simply compare before and after?

From time to time you will hear managers say that the effectiveness of marketing can be determined by simply comparing two market results, i.e. sales, revenue, or market sales that come from two separate budgets. This produces something like a before-and-after analysis. However, relying on this would be careless, as the result may have been distorted by the influence of third-party factors. Here are two examples: compare an ad campaign for chocolate from July to September with one from October to December. The poorer distribution during the summer months could have potentially influenced the final result. The second example would be to compare the deployment of a company’s sales force in the last quarter of the previous year with the first quarter of the new year. In this case, sales in the first quarter could be negatively influenced by the fact that customers tend to decide on purchasing budgets towards the end of the previous year.  

3. Data Analytics Help with Well-Established Products

Analyzing data typically obtained from market research institutions presents one possibility. These data usually provide sales and revenue per month or quarter and the respective budgets used for advertising, the use of field sales forces, etc. for all competitors in a market. Now, with the help of data analysts, it is possible to statistically determine a functional relationship that best explains the variation of sales in relation to budgets. However, this approach only works for well-established products with a sufficiently long time-series. Only then can this function be derived empirically. The assumption here is that the effectiveness will continue to behave more or less as it did in the past. See the example shown in (6).

4. Results Can Only Be Used If a Causal Relation Can Be Assumed

Empirical data analysis usually specifies that the sales of a product depend on the use of marketing activities. In the case of advertising, it is possible to work with budgets, provided that they have been recorded. However, it is also possible to work with variables that describe the use of these activities in more concrete terms, e.g. Gross Rating Points (GRP) for newspaper advertising, i.e. the percentage of the target group reached by newspaper advertising multiplied by the number of placements. That said, it is not always clear as to whether sales are causally linked to advertising. Rather, some companies determine their advertising budget as a percentage of the most recently generated sales. This then produces reverse causation, and no conclusion can be drawn about the effectiveness of the advertising. In another example, the author was dealing with a cold remedy where the company advertised four times as much in winter as  in summer. In this case, the effectiveness cannot be deduced from the variation of the past data regarding revenue and advertising. Here it proved beneficial to look at the variation of market shares and advertising shares in the relevant overall market.

5. The Assumed Form of the Relationship Needs to be Optimizable

Data analysts cannot simply work with linear functions, as is usually the case, because then the optimal budget derived from them would either be a) zero if the increase in the contribution margin resulting from the sales is smaller than the budget required for it, or b) infinitely large should the corresponding effectiveness be greater than the used budget. In a linear relationship, there are no diminishing marginal effects that would lead to an optimum. In this respect, it is important in marketing to work with non-linear functions, i.e. curves, from which an optimum can be derived (2). Often it is useful to convert the variables of the advertising effort into logarithms in order to reach decreasing marginal effects for increasing advertising expenses.

6. Do not Forget About Carry-Over Effects in the Relationship

When setting up the basic response function, it should not be forgotten that sales do not decline in the same proportion when advertising budgets are cut. On the contrary, all previous marketing expenditures have led to brand equity, which for a while leads to sales that are only marginally decreasing. However, you should keep in mind that although you are slowly losing sales, it takes just as long to rebuild them. This is known as the carry-over effect. Empirically speaking, this has been found to be around 60%, i.e. 60% of the past advertising is recalled (5). This can be explicitly taken into account by including past sales as an explanatory variable. In the same way, past marketing expenditures should be used as an explanation. Sometimes there are also delayed effects. For example, investments in research and development only lead to sales in the long term once the developed product has been launched on the market. Respectively, the long-term effects should be taken into account when determining the budget. In our example, these are 2.5 times the short-term effects.
1/(1-carry-over) = 1/(1-0.6) = 2.5

7. Goodness of Fit of the explanatory power is not enough! You have to evaluate the forecast error.

It is often found that data analysts choose the functional relationship that ensures the best fit. This means that weights or parameter values are determined for the variables that maximize the explained variance. However, this only acknowledges the quality of the explanation, and not whether the model with the functional relationship can be used for predictions. For this purpose, what is known as cross validation is recommended, in which the functional relationship is estimated on the basis of part of the data (e.g. 75%) and then applied to the remaining 25% to determine the forecast error. Only then is it possible to know how well the model can be implemented.

8. Marketing Budgets Are Proportional to Elasticities  

A good indicator for describing the effectiveness of marketing instruments are elasticities (2). They describe by how much the percentage of sales of a product change if, for example, the advertising budget is changed by a certain percentage. Then it is only necessary to divide the relative change in sales by the relative change in the advertising budget.

Advertising elasticity = Relative change of sales (%) / Relative change of advertising budget (%)

The good thing about this key figure is that it is dimensionless and can be compared across all products and marketing instruments. Furthermore, this allows the optimal budget to be determined depending on the elasticity (3). The following applies in the optimum:

Budget = profit contribution x margin x elasticity

If you apply this to all your marketing instruments (except price), then the budgets of the different instruments should be proportional to their respective elasticity. It should be noted that you should use the long-term elasticity, which is typically calculated by dividing it by 1 - carry-over (see point 6).

Conversely, this formula can also be used to calculate what elasticity has been assumed in a marketing plan. Furthermore, you can calculate what the optimal percentage of the budget should be of the revenue, namely that which corresponds to the elasticity multiplied by the margin. If, for example, an advertising elasticity of 4% applies and a company has direct costs of 75%, which results in a margin of 25%, then an optimal advertising budget of is 4% x 25% = 1% of sales. This is, for example, what most car manufacturers spend. It does not sound like much, but when extrapolated onto VW’s total rvenue of 253 billion euros, the result is an advertising budget of 2.5 billion euros.

9. New Products Need Analogical Conclusions

Data analysts can only be employed if historical data can be used to draw conclusions about effectiveness, which is only possible for existing products. For new products, it is necessary to work with analogical conclusions. One way of doing this is to use the effectiveness of a marketing activity for the product most similar to the new product. However, if such analyses are not available, it is recommended to refer back to evidence that has been published in professional journals. Such journals now provide meta-analyses of all individual analyses that have ever been published in scientific journals, and it is always easy to produce a summary of the results that have been published so far when elasticities can be used, or if the results can be converted into elasticities. In these meta-analyses of elasticities for different marketing activities, the mean values and frequency distributions of the elasticities are given. From these analyses it is known, for example, that the advertising elasticity is 10% (tending towards decreasing) (4). For the sales force it is about 30%, which is three times as high as it is for advertising (1). In addition, the factors which the differences in the individual results depend on are also indicated. This allows the mean value to be corrected for the influence of the product type, geographical region, or methodological details, e.g. whether the elasticities relate to sales, revenue, or market share. In this way, you gain at least a plausible value for the elasticity, which can be compared with your own experience, but which also provides food for thought if you want to work with completely different assumptions. In any case, with the assumed elasticity, as shown in point 8, the optimal budget can be easily determined

10. Experiments Are Best

The methods shown so far are of course subject to some limitations. Elasticities derived from historical data are, in principle, only valid for the period under investigation. Whether they will also apply in the future is difficult to predict. In general, it is certainly true that the more stable the market, the more elasticities from historical data make sense, and, conversely, the more dynamic the market is, the less elasticities make sense. To be sure, results from meta-analyses can provide plausible values, but nothing more. Carrying out field experiments is always the most appropriate approach, because this is where you can observe the real effect.

In the field of the Internet, what is referred to as A/B testing has become established for precisely this purpose. The advantages lie in its direct application in web design. But even in the offline world, it is always possible to carry out smaller experiments, e.g. suspending advertising in regional daily newspapers in some regions in order to be able to determine the effect in comparison to the other regions that have acted as a control group. It is also possible to test different visit frequencies of sales people over a shorter period of time in the sales field.

So, you can see that it is unnecessary to merely rely on your gut feeling when setting marketing budgets. Rather, data analyses, meta-analyses, and experiments offer sufficient opportunities to proceed taking an evidence-based approach.