ROI Analysis of Pharmaceutical
An Independent Study
! Measure ROI for detailing (DET), direct-to-consumer
advertising (DTC), medical journal advertising (JAD),and physician meetings & events (PME)
! Understand how ROI differs according to brand size (in
The objectives of this study are fairly ambitious: first, to measure the ROI fordetailing (DET), direct-to-consumer advertising (DTC), medical journaladvertising (JAD), and physician meetings and events (PME). The secondobjective is to understand how the ROI changes according to the brand sizeand launch date. This is a unique study in which we look at all four activitieson the same footing, and ask the question, “What is the productivity of thesemarketing expenditures?” While this endeavor is most challenging from astatistical standpoint, it is also highly relevant for those in the pharmaceuticalindustry who are determining the marketing mix.
! Analyze using standard statistical techniques (ordinary
– Measure ROI for median brand profile– Measure ROI for median brand profile within
There are three components of our overall approach: (1) We use historicaldata; (2) we analyze these data using standard statistical techniques; and(3) we do the analysis in aggregate.
We’ve assembled a comprehensive historical database of branded productsover an extended time period. What we analyze is what has actuallyhappened in the marketplace. It’s not an experiment, it’s not a test. Thebenefit of this is broad generality…as well as realism. While historical dataanalysis has its strong points, it is only one approach to measuring ROI —there are many other approaches as well. All of these approaches shouldbe used together. As noted, the benefits of this approach are generality andrealism. However, there is a price to pay for that realism — namely, thingsget complicated. And thus, we have to use a fairly sophisticated statisticaltechnique known as ordinary least squares regression. This is a standardstatistical method. So while the situation is challenging, there is a goodstandard statistical technique for approaching it.
The third component of our approach is the notion of an aggregate analysis.
What we’re going to be doing is taking advantage of the breadth and realismof our data to come up with an ROI measure for an “average brand.” We willalso further calculate the ROI for an average brand within specific size andlaunch date cells. We will represent the average brand by a "median brandprofile," which I will soon describe.
! Includes all brands with ≥$25MM in revenues in 1999
The data used included all brands with more than $25 million in revenues in1999…for a total of 391 branded drugs. Also in the database are 127generics. The data evaluated cover the period from 1995 to 1999, a 5-yearspan. This provides us with 16,696 total monthly observations (391 brandeddrugs times an average of 43 months of data for each brand).
detailing dollars: includes 1-to-1 physicianvisits, as well as rep-driven small group meetings& events (Scott-Levin)
direct-to-consumer dollars: includes television,radio, print, and outdoor (Scott-Levin)
medical journal advertising dollars (PERQ/HCI)
physician meetings & events dollars (Scott-Levin)
number of scripts filled at retail (Scott-Levin)
retail pharmacy price per script (Scott-Levin)
Key variables used*: Detailing (DET) was defined as detailing dollarsincluding 1-to-1 physician visits, as well as rep-driven small group meetingsand events (data provided by Scott-Levin). Direct-to-consumer (DTC)advertising included television, radio, print, and outdoor (data provided byScott-Levin). Journal advertising (JAD) comprised both primary care andspecialty journals (data provided by PERQ/HCI). Physician meetings andevents (PME) covered physician meetings of many kinds, but excluded rep-driven small group meetings and events (data provided by Scott-Levin).
Those are the independent marketing activity variables. Scripts refer to thenumber of prescriptions filled at the retail pharmacy level (data provided byScott-Levin). And finally, to get return on investment in terms of dollarrevenues, we have price per script at the retail pharmacy level (dataprovided by Scott-Levin).
*See Definition of Terms on this site.
Branded Drug Marketing Expenditures
Total Branded Marketing Expenditures grew at 16.2% per year
This slide shows the progression of spending on these four marketingactivities over time. Detailing is the dominant expenditure, and has beenincreasing over time (over 10% per year). In 1995, over $3 billion werespent on detailing, and it’s up to almost $5 billion by 1999. When one looksat JAD, DTC, and PME, it is apparent that one of these activities is movingahead pretty strongly, with a growth rate of over 50% per year…and that’sDTC. DTC grew from $200 million back in 1995 to a billion and a half dollarsin 1999. PME has been growing also, although not quite as fast (+23.6% ayear). Journal advertising has lagged, with just 9.8% growth per year. Whilethere was an overall growth for JAD between 1995 and 1999, that’s mainlybecause the 1995 base was pretty small. In fact, JAD expenditures crestedin 1997 — it’s clearly lagging behind the others in terms of growth.
! Regression analysis examines all the brands in all the
months in which they were marketed. It then analyzeshow changes in expenditures over time correlate withchanges in script levels over time. Regression willconclude that a marketing variable has a high ROI if thatvariable can consistently explain changes in script levels
! Regression analysis controls for causes of sales not
attributed to a brand’s marketing (eg, genericpresence/marketing, external trends affecting brandgrowth, competitive spending, price)
Regression analysis examines all the brands in all the months in which theywere marketed. It then analyzes how changes in expenditures over timecorrelate with changes in script levels over time. Regression will concludethat a marketing variable has a high ROI if that variable can consistentlyexplain changes in script levels over time.
What we start with is a spreadsheet with 16,000 rows in it, each rowcorresponding to a different brand for a different month. If one looks downthat spreadsheet, one would notice that scripts went up in a particular monthfor a particular product. Why did that happen? To explain this, one wouldthen look at the expenditures for DTC, detailing, journal advertising, andmeetings and events. If, for example, only detailing expenditures went up ina month in which scripts also went up, one might conclude that detailingexplained the increase in sales, and thus it had a high ROI. But, that’s justlooking at one row of data; regression looks at 16,000 rows, notes whenscript sales go up or down, and then tries to explain those changes in termsof changes that have occurred in all the marketing variables. To the extentto which a given marketing variable can explain those month-to-monthchanges, we say it’s a powerful explainer of sales. It has a high ROI.
However, it’s also important to control for other things that occur in themarketplace. Regression analysis can control for causes of sales notattributed to a brand’s marketing: these include generic presence andmarketing; external trends affecting brand growth (like growth in the categorythat’s due to a growing awareness of a disease among physicians orpatients); competitive spending (for example, a bigger brand marketing effortmay affect scripts not only for that brand, but also for its competition); andprice.
! ROI (Return On Investment) = Increase in revenues per
! The % Increase in Unit Sales depends on brand size,
launch date, PCP fraction, and size of marketingexpenditures (DET, DTC, JAD, and PME)
ROI, or return on investment, is defined as the increase in revenues peradditional dollar spent. But, how do we actually compute that ROI? Whatthe regression literally tells us is by what percentage units will increase perdollar spent. Once that percentage increase is known, it is multiplied by thebase unit sales, providing the total number of units by which that dollar hasincreased sales. This is then multiplied by the price, to get the totalrevenues generated per dollar expenditure.
It should be emphasized here that we’re calculating the effect of themarginal dollar change. What it can tell you is that in general, over this 5-year period of time, here’s what happened because of the extra dollars thatbrands have spent from month to month…and therefore, here’s the ROI onthose investments.
The other thing that should be kept in mind is that the percent increase inunit sales will differ significantly by brand. The percent increase in unit saleswill depend on brand size, launch date, fraction of sales that go to primarycare physicians (PCP), and finally, level of marketing expenditures (DET,DTC, JAD, and PME).
Median Brand Profile
! The data were highly skewed by brands with very high
! The skewness makes the mean
not representative of
! We therefore defined average brand profile by taking the
levels of price, base sales, and marketingexpenditures
! Using the mean would generally increase ROIs fairly
In doing an aggregate analysis, we try to come up with the ROI for theaverage brand. That begs the question, “How do you define the averagebrand?” The data were highly skewed by brands with very high prices, saleslevels, and expenditures. Most of the brands were in one region in terms ofprice and expenditure level; however, there were also strong outliers. Whenone is in that situation, the skewness makes the mean
not representative ofthe average brand, or where there is the bulk of the data. A commonremedy is to use the median
, literally the data point with an equal numberabove and below. We therefore defined an average brand profile by takingthe median levels of price, base sales, and marketing expenditures, andcalculated the ROI if that brand increased expenditures by one dollar. If wehad used the mean, all of the numbers would have increased fairlyuniformly.
Revenue/Launch Year Cells
Numbers Represent Median Brand Profile in Each Cell
This slide shows the median brand profiles broken down into nine cells.
There are three levels of brand size ($25-50 million, $50-200 million, and$200 million and above), and three levels of launch date (before or equal to1993, 1994 to 1996, and 1997 to 1999). What you first see in each cell isthe number of brands in each of these categories. For example, in the $25-$50 million category launched in 1993 or before, we have 66 brands. Also,what you see is that the majority of brands analyzed were launched before1993. However, there are a sufficiently large number of brands launchedbetween 1997 and 1999. The primary care physician (PCP) fraction showsthat roughly 20% to 50% of sales come from primary care physicians for thesmaller brands, but for the very large brands, there is a higher PCPpercentage (in the 50% to 60% range). Price per script, scripts per month,and revenues per year for each cell are also shown here.
Revenue/Launch Year Cells
Numbers Represent Median Brand Profile in Each Cell
This slide is also broken down into the same nine cells, but provides theexpenditure by marketing variable for a median brand profile in each cell.
Not surprisingly, expenditures tended to be larger for more recent and largerbrands. However, one thing that might be questioned when looking at thesedata is that there are zeroes for some median expenditures, especially DTC.
That means the median brand wasn’t using DTC in a particular month.
While that median was zero, we do have a fair amount of variation aroundthat zero, and that’s what the next slide shows.
Brands in $200MM+, Launch Year 1997-1999 Cell
! Although the median may be zero for DTC, we have data on various levels of
Here we have the range of monthly expenditures for the products with over$200 million in sales, and launched between 1997 and 1999 (functionally,the lower right-hand cell of the previous slide). While the median monthlyDTC level was zero for these brands, for almost every one, there’s datashowing high monthly expenditures. Just because the median was zero, itdoesn’t mean that there is no right to say something about the effectivenessof the variable. As long as there is a wide range of expenditure, theregression analysis is able to identify instances where DTC increased, andto see how well this explained changes in sales.
These are the ROIs for each tactic for the overall median brand, whichturned out to be a brand launched at or before 1993, in the $50 to $200million range. For that overall median brand, the ROI was $1.72 fordetailing, $.19 for DTC, $5.00 for journal advertising, and $3.56 for physicianmeetings and events. Note the margin of error…meaning that there is 95percent confidence that the detailing ROI is somewhere between $1.53 and$1.91. It is still generating a significant return on investment. When onelooks at JAD and PME, their ROI numbers are noticeably higher. If oneremembers the expenditure levels shown previously, especially for journaladvertising but also for PME, these are obviously underutilized resources.
You know that physicians do not see millions of journal ads for any givendrug. They may have seen a few, but when you add an additional one, it'smore likely to have an effect.
Now, what about DTC? It’s conspicuous at 19 cents. But here again, there’sa pretty large margin of error around that level (19 cents plus or minus 52cents). That means, statistically, one can’t distinguish the effect of DTCfrom zero because that 95% confidence interval goes from a negative to apositive. Why is DTC coming out as less effective? Your conjectures are asgood as mine. Is DTC that targeted? What percentage of the audience is itrelevant for? While with DTC we can focus promotion to some extent (forexample, we can advertise during certain shows that older people watch), itstill is not as targeted as the other tactics. Another potential answer is that1995-1999 was probably a period of learning with DTC. Firms are stillinvestigating which type of creative execution works best.
! PME ROI has a large margin of error (Plus or Minus
$1.92). It is highly correlated with other marketingvariables (especially DET), making it particularly difficultto determine PME ROI by size/launch date
There is one additional point to note for PME. While it has a significant ROI,there is also a high margin of error ($1.92). What drives this larger margin oferror for PME is that PME is very highly correlated with other marketingvariables, thus making it very difficult to get a good fix on the exact ROI forthis tactic. The regression analysis may find that when scripts go up in agiven month, PME expenditures also go up, suggesting a high ROI for PME.
However, often when PME expenditure goes up, something else will also goup at the same time, especially detailing. When that happens, it is moredifficult to calculate the ROI contribution for PME, and that’s why there is alarge margin of error. This became an even greater problem when the datawere further split by size and launch date, and so we won't be showing thePME splits for the nine cells. The following slides will, however, provide theROI for detailing, DTC, and journal advertising broken down by revenue andlaunch year.
ROI of Median Brand Profile by Size/Launch Date
n.s. = not statistically different from zero, two-tailed test, 0.05 significance level
This slide shows the ROI results for detailing, DTC, and journal advertising,broken down into our nine cells. The first thing one is struck by is that theROI is higher for larger brands. For detailing, in addition to the ROI beinghigher for the larger brands, it’s particularly high for the more recent brands.
This study would support the adage that when you’re launching a newbrand, get that push strategy going, and get the people in the field detailingphysicians and pushing it through. That expenditure can have a hugepayoff…$10.29. Now, let’s look at DTC. The "n.s." means not statisticallysignificant. We could not detect a statistically positive effect for DTC incertain cells, but could for the more recent brands. The ROI for the medianbrand profile launched between 1997 and 1999, with at least $200 million insales, was $1.37. In general, DTC does better for larger and more recentlylaunched brands. Journal advertising follows that same pattern in terms ofhaving a larger ROI for the larger brands. However, contrary to detailing andDTC, journal advertising tends to be even better for the mature brands. But,it’s not a drastic difference, as JAD does have a high ROI for recentlylaunched brands as well.
Long-Term Effect of Marketing
! Total effect of promotion is not just realized in the first
month — it accrues over several months. At most,50% of the total ROI is felt in the first month — can take1-2 years for most of the impact to be realized
! Half of the total ROI payback for DET and JAD occurs
immediately in the month in which the expenditure takesplace. It takes 10 months for the full effect to accumulate
Cum ulative DET ROI
Cum ulative JAD ROI
This slide provides an interesting perspective on what we call the “dynamics”of advertising effectiveness. That is, the effectiveness of a marketingexpenditure is not only seen in the month that you spend it, but it alsoaccrues in subsequent months. In fact, for our variables, at most 50% of thetotal ROI occurs in the first month. It can take 1-2 years for most of theimpact to be realized. With detailing and journal advertising, exactly half ofthe ultimate ROI impact occurs in the first month; the other half occurs in thesubsequent 9 months. For example with detailing, the total detailing ROIwas $1.72. Half of it ($.86) was felt immediately in the first month.
However, it then took up to a full 9 months for the total ROI of $1.72 to beaccumulated. In contrast, for DTC, only 10% of the effect takes place in thefirst month, 72% in the first year, and then it takes over 2 years for the totaleffect to accumulate.
Allocating New Marketing Funds to DET vs DTC
Median Brand Profile: $50-200MM, Launched 1994-1996
! What is the ROI for various allocations of a 15% increase
! Allocating additional funds to DTC, which in this case is less
productive than DET, dilutes ROI of budget increase
What I would like to do now is take you through budget scenarios showingwhat might be the implications of these findings. Here we’re going to look atthe median brand profile of a $50 to $200 million brand, launched between1994 and 1996. The ROI was $2.68 for detailing and 43 cents for DTC.
What we’re going to look at is a scenario where the brand is expanding itsmarketing. What would be the implications of putting the additional moneyinto detailing versus DTC? How does that affect the ROI payback? For themedian brand profile in this cell, the current detailing and DTC combinedbudget is $5.256 million. What would be the return on investment for variousallocations of a 15% ($788,000) increase in the current budget? What arethe implications of putting that increase into detailing versus DTC? The firstline in the chart shows the current state of affairs. We’re spending $5.256million on detailing, but nothing at this point on DTC. Thus, the total budgetis $5.256 million. Our annual revenues are $59.784 million. If another$788,000 were added to the marketing budget, what would happen if we putall of this into detailing? The regression analysis predicts that the annualrevenue would go up to almost $62 million, a gain of over $2 million, on a$788,000 expenditure. This means that the ROI on that budget increase is$2.71. What would happen if we didn’t put it all into detailing, but put halfinto detailing and half into DTC. That’s going to dilute the ROI on the budgetincrease (now $1.54) because you’re putting some of the promotionalincrease into something that’s less productive, in this case, DTC. You’realso going to be losing potential revenue. What would happen if weallocated all of the increase to DTC? In this case, the ROI would decreaseto 39 cents.
Allocating New Marketing Funds to DET vs JAD
Median Brand Profile: $200MM+, Launched ≤1993
! What is the ROI for various allocations of a 15% increase
! Allocating additional funds to JAD, which in this case is more
productive than DET, enhances ROI of budget increase
In this scenario, we will use a 15% budget increase, but allocate thepromotional increase between detailing and journal advertising. We will alsonow look at a larger brand — greater than or equal to $200 million in annualrevenue — launched on or before 1993. For this median brand, the detailingROI is $2.34 and the journal advertising ROI is $6.79. You can see that inthis case, it’s going to be better to allocate more money not to detailing, butto journal advertising. Here we’re spending $16.128 million on detailing andonly $624,000 on journal advertising, for a total budget of $16.752 million,and annual revenues of $461.139 million. Now let’s take that 15% budgetincrease, which in this case would be $2.513 million, and put it all intodetailing. You do get some payback for that because detailing does have adecent ROI. In fact, the ROI on that budget increase would be $2.43. If youdivide the budget increase, however, between detailing and journaladvertising, you now would get an ROI on that increase of $4.50. If you putall of the increase toward journal advertising, then you get an ROI of $7.06.
Obviously, putting all your money into the more productive activity, in thiscase journal advertising, provides the highest ROI.
Reallocating Budget Between DET and JAD
Median Brand Profile: $200MM+, Launched ≤1993
! What is the ROI when one reallocates the current $16.752MM
! Reallocating more toward JAD, which in this case is more
†Base Revenues = Estimated Brand Revenues if no money spent on DET or JAD (ie, if Total Budget = 0)
This scenario takes the same median brand profile seen in the previouscase, and looks at what happens when one reallocates the current budgetbetween detailing and journal advertising. The detailing ROI is $2.34, andjournal advertising is $6.79. But what would happen if we started to takesome of that detailing money and put it toward journal advertising? Howwould that impact the ROI of the total marketing budget? What we start offwith is the current budget, where we are spending $16.128 million ondetailing and $624,000 on journal advertising, for a total budget of $16.752million; annual revenues are $461.139 million. Next we calculate a baselevel of sales, which is an estimate for the sales levels you obtain withoutany marketing budget. For this median brand, the base revenues are a littleover $416 million. This means that the marketing budget ROI is now $2.66($461.139 million minus $416.518 million divided by $16.752 million). Now,let’s start reallocating some money to journal advertising. What you see isthat the ROI of your marketing budget obviously improves. But in thisexample, it’s not highly noticeable until journal advertising dollars areincreased significantly. When you move from a journal expenditure of$624,000 to $1.248 million (an increase of over $600,000), it’s still a smallpercentage of the total budget, and thus you are not going to see a hugeincrease in the ROI. Allocating more money to journal advertising, in thisnext case close to a $2 million increase, then one starts to see a morenoticeable effect with the ROI going to $3.26; for over a $4 million increase,the ROI then goes to $4.32. What comes out of this study is that if you aregoing to do some reallocation, reallocate to the higher ROI tactic (journaladvertising in this example), to make the overall marketing budget moreproductive.
! Overall ROI of $1.72 suggests that DET pays off even at very
! Range of $1.27-$10.29, depending on brand size/launch date
! Particularly higher ROI for large and more recently launched
! Overall ROI of $0.19 suggests one must be careful to use DTC
! Range of $0.00-$1.37, depending on brand size/launch date
! Best for large and more recently launched brands
Our overall findings: for detailing, there is an overall ROI of $1.72,suggesting that detailing pays off even at very high levels of expenditure.
The range of this variable is the largest, going from $1.27 to $10.29,depending on brand size and launch date. There is a particularly high ROIfor large and more recently launched brands. For direct-to-consumeradvertising, the overall ROI is 19 cents, suggesting that one must be carefulto use DTC for the right brand. DTC doesn’t appear to be as simple as “justthrow money at it and good things happen” — that’s an important lesson tolearn. However, we did find that the large and more recently launchedbrands (showing an ROI of $1.37) are typically the more promising areas.
Overall Findings (cont'd)
! Overall ROI of $5.00 is highest among all four marketing activities
! This plus its small share of budget suggests JAD is underutilized
! Range of $2.22-$6.86, depending on brand size/launch date
! Particularly higher ROI for larger and older brands
! This plus its small share of budget suggests PME is underutilized
In terms of medical journal advertising, here we have an overall ROI of $5,which is the highest among all four of the marketing activities. This finding,plus its small share of budget, suggests that journal advertising isunderutilized, and may be an area where firms need to focus a little bit moreattention. The range calculated was $2.22 up to $6.86, depending on brandsize and launch date, with particularly higher ROI for larger and olderbrands. Physician meetings and events carries a very similar message tothat of journal advertising. The overall ROI of $3.56 is second highest. Thisfinding, plus its small share of budget, suggests that PME is alsounderutilized.
Overall Findings (cont'd)
! Up to half of the ultimate ROI from a marketing
investment occurs in the first month of expenditure. Cantake 1-2 years for most of the impact to be realized
! ROI from incremental marketing investments or
reallocations depends significantly on which activitiesget additional monies
In terms of other lessons, there is the finding that the total ROI takes placeover time: at most, half of the ultimate ROI from a marketing investmentoccurs in the first month of expenditure, and then it can take up to 1-2 yearsfor most of the impact to be realized. A final lesson is that ROI fromincremental marketing investments or reallocations depends significantly onwhich activities get additional monies. There are differences from onemarketing activity to another, and how budgets are allocated can make a bigdifference in the overall productivity and returns from marketingexpenditures.
Additional Analysis/Next Steps
– Very challenging because of smaller sample size– Maximum of 60 months versus 16,696 monthly
– More observations– More recent ROIs
– TV versus print– Recent versus earlier ROIs
While we’ve learned a lot from this analysis, it is time to plan for futureanalyses. One important next step would be to calculate the ROI for specificbrands, rather than for the “median brand profile.” The challenge in doingthis is that there are only, at most, 60 months for each brand. This is arelatively small sample size, especially compared with the 16,696 monthlyobservations we have for the Aggregate Analysis. One way to help addressthe issue of limited data.is to collect more data. And that is why we willsoon be adding the year 2000 data to our database. In addition to helpingthe brand-specific analysis, this will provide more recent, and thus morerelevant, ROIs. We also plan to explore the competitive impact of onebrand’s expenditure upon others in its therapeutic category. And finally, wehope to examine DTC in more detail. In particular, we’ll investigatetelevision versus print expenditures, and recent versus earlier ROIs.
Clearly there is still much exciting analysis to be done. We look forward tobringing you these additional results in the future.
Café Gratitude is our expression of a world of plenty. Our food and people are a celebration of our aliveness. We select the finest organic ingredients to honor the earth and ourselves, as we are one and the same. We support local farmers, sustainable agriculture, and environmentally-friendly products. Our food is prepared with love. We invite you to step ins
SEASONED WOOD AND TIMBER STRUCTURES CARPENTER ANTS Preventive Control - Carpenter ant colonies may establish in stumps, unhealthy or dead trees, and in various interior andexterior buildings; wood structures such as studs, rafters, baseboards, door casings and othersupport structures. Exterior decaying structures are particularly vulnerable and should be replaced. Outside stacked wood mat