Welcome to Analysis of ROI for Pharmaceutical Promotion

Direct-to-Consumer Advertising (DTC)

16. Why is it that DTC has such a low ROI compared to other activities?
There are many possible explanations:

•	DTC is not as easy to target since it is primarily done over mass media
•	A DTC ad affects the medication of only one patient, whereas DET, JAD, and PME affect the prescribing behavior of one physician, who has several patients. These activities thus have a multiplier effect that DTC does not have.
•	DTC advertising strategy has changed over time as companies are learning how to use it. They've changed creative strategy, message strategy, and scheduling strategy, and are in the process of learning how best to do DTC. The other activities have been around for a long time, so companies better know how to optimize them. Thus, whatever effect we read out for DET, JAD, and PME probably represents the best that companies can do, whereas the DTC results probably reflect a lot of campaigns that were far from optimal.

17. Do these results mean that DTC is generally ineffective?
It is true that in this study, DTC's ROI was not statistically significant for the overall median brand. However, based upon this information, one should not make a sweeping statement that DTC is "generally ineffective" for the following reasons:

1.	The overall median brand profile in the study was for a mature brand in the $50-$200M range. While this brand profile is the typical brand for our overall data, it is not the typical brand that uses a lot of DTC. For the recently-launched large brands ($200MM+), we found a statistically significant ROI of $1.37.
2.	As pointed out in the answer to Question 16, 1995-1999 was a time of learning for the industry with respect to DTC, and a period of change in the regulatory environment. Thus, it is possible that the ROI for DTC has increased over time. This was something we were not able to explicitly test, but hope to examine in the future as more data is incorporated.

Overall, perhaps the most appropriate conclusion is that generating incremental revenues from DTC is not as simple as just "throwing money" at this tactic. The indications seen are that it may not be appropriate for all brands, and that additional learning needs to accumulate.

18. Is there a "threshold" below which DTC advertising isn't a good investment? How might this impact the results?
The analysis did not explicitly measure threshold or saturation levels. However, if there is a threshold effect, this might explain the statistically insignificant ROI for the average overall brand, which is a $50-$200 million brand launched before 1994. Brands with this profile might not be investing in DTC beyond the threshold level.

19. Can DTC be broken out into direct-to-patient?
DTC is sometimes called direct-to-patient. However, if by direct-to-patient one means mailing information directly to patients, then no, the Scott-Levin DTC audit does not include this type of promotion. For clarification, the Scott-Levin DTC data are supplied by Competitive Media Reporting and can only be broken out among 11 media types: Network TV, Spot TV, Cable TV, Syndicated TV, Sunday Magazines, Consumer Magazines, National Spot Radio, Network Radio, National Newspapers, Local Newspapers, and Outdoor Advertising.

20. Do you have information on the ROI of Direct Mail to Consumers? This was not in your definition of DTC.
Direct Mail is not included in the DTC definition. (See Question 19 for more details.)

21. Please explain why DTC is shown to have a higher ROI for larger brands; ROI is return on $1 spent, so it should be indexed. But you mentioned that large brands spend more. This shouldn't matter; it's the $ return for $1 spent, right?
The regression estimates ROI by examining how much sales increase per dollar increase in advertising. The higher DTC ROI for larger brands simply means that the incremental dollar spent on DTC for a larger brand usually generated more sales than an incremental dollar spent for a smaller brand. It is true that larger brands are as a rule spending more on DTC than smaller brands, so the incremental changes in expenditures the regression observes for larger brands are generally off of a larger base. It could be that there is a base threshold before DTC has any effect, and small brands are just not above that threshold.

22. How can the margin of error for DTC (+/- $.52) be greater than the ROI ($0.19)?
It is quite possible, and not uncommon in statistical analysis of advertising, that the margin of error is greater than the computed effect. In this case, the return for DTC could be anywhere between minus $0.33 and plus $0.71. This is the way that regression tells us that it is not 95% confident that the return for DTC exceeds zero. Note, this general approach to margins of error and "confidence intervals" is common to all statistics, not just regression, and certainly not just to this study. In statistical terms, the overall DTC ROI is not statistically different from zero.

23. The DTC ROI was $1.37 for the larger, more recently launched brands (greater than $200 million, launched 1997-1999). What was the margin of error?
It was +/- $1.12.

24. Is it possible that the low ROI on DTC is due to the fact that DTC expenditures sometimes swing wildly from month to month, but sales don't tend to do so (and so it is not a good predictor of changes in sales, and thus gets a poor coefficient in the regression model)?
Yes, that is possible, but remember that the regression also considered that DTC might increase sales in future months. So it is possible, and in fact this was the general finding, that only about 10% of DTC's total effect is felt in the first month in which the money is spent. It took about 2 years for virtually all of DTC's effect to accumulate.

25. If DTC takes over 2 years for the full effect, has this all been accounted for in your ROI calculations for the recently launched brands?
We have not cut short the effect in future periods. The 2-year carryover is included in the 1997-1999 ROI calculations.

26. If DTC is effective on an accumulated basis over time, and given the possible few-month lag between when patients see an advertisement and when they finally fill their prescription, is the monthly correlation of change for the two variables the right way to measure the impact?
The regression considers the effect not only in the first month but in subsequent months as well. So it looks not only at the correlation between current month expenditure and current month sales, but at future sales as well.

<<prev	Jump to other FAQ's	next>>
Study Independence \| Detailing (DET) \| Direct-to-Consumer Advertising (DTC) \| Medical Journal Advertising (JAD) \| Physician Meetings & Events (PME) \| Timing Effects \| Return on Investment (ROI) \| Regression Model \| Miscellaneous