Do the Segments Stack Up?
Market segmentation was developed by package goods marketers to deal with two problems: consumer demographic and purchasing information that is available for groups but rarely for individuals, and advertising media that can target groups reasonably well but individuals almost not at all. Market segmentation is not the end all be all; it is a way to deal with sketchy information and scatter-gun media.
For the package goods giants, market segmentation is obviously superior to spreading marketing resources randomly. It enables them to spend less money advertising to low potential consumer groups and to heavy-up against groups that appear to have the greatest potential to respond. However, there is still enormous unavoidable waste and missed opportunities.
Things are different for pharma companies. They know almost exactly how much prescribing each doctor has been doing. Their sales reps can choose to call on individual doctors as often as they will allow.
Pharmaceutical companies could treat each doctor as a unique market but they instead deal with individuals as members of market segments. By emulating the package goods giants, pharmaceutical companies behave as if they don’t know as much as they do about their doctors and cannot target details and samples any better than P&G can target TV commercials.
The arrival of doctor-level prescribing data years ago should have changed pharmaceutical marketing much more than it has. Breakthrough analytical tools make it possible to take this data and quantify each doctor’s unique responsiveness to details and samples for a brand. This means that marketing resources can potentially be allocated much more profitably.
Instead, the data are used as just another segmentation tool. It’s like limiting a great race horse to Sunday rides around the park.
Within the industry, segmentation results in valuable sales calls going to doctors who are extremely unlikely to write more scripts as a result. PBE’s analysis shows that this is the case for about 40% of sales calls. Although some of this waste can be blamed on sales reps going off plan, we’ve seen many examples where reps have actually done better by cheating on the plan.
It’s not as if the wasted calls couldn’t have been put to better use. Frequently, wasted calls could have been used to address profitable opportunities that market segmentation failed to identify.
For example, one company found that the average doctor who was identified as highly responsive to detailing (using the breakthrough methodology) but was not included in targeted segments wrote three times as many incremental scripts per sales call as doctors who were targeted through segmentation but did not make the responsiveness cut. This example, by the way, involved a situation where reps went out and did their own prospecting.
Market segmentation dramatically reduces sales force productivity in comparison to a new individualized approach. This approach, Call Value Targeting, looks at each doctor as a unique market and each potential sales call as a unique investment opportunity with a definable value. It overcomes the flawed assumption in segmentation that all individuals in a segment are the same.
Call Value Targeting also addresses a subtle, but extremely damaging, flaw in the way market segments are typically constructed. Doctors are usually targeted according to their decile of past script writing. Each decile is assigned a call frequency with heavier writing deciles being targeted for more calls than lighter writers.
Marketers focus on the volume of past prescribing when segmenting doctors and allocating resources because…“It’s better to increase the share of a heavy writer than a lighter one.” This logic would be impeccable, if one observed a given amount of detailing producing about the same share change across the board. In fact, one observes considerable variation.
In the real world, individual doctor’s responsiveness to detailing varies so much that doctors who are not heavy writers of a brand or category may be considerably more responsive to detailing of a brand than those who are. Quantifying the variation in individual doctor’s responsiveness to future detailing is the key to uncovering significant profits that are consistently left on the table.
The Call Value Targeting approach goes beyond simply targeting individual doctors. It targets individual call opportunities (which happen to be “attached” to individual docs) based on the number of incremental scripts that will most likely result from making any given call.
Call Value Targeting can be far easier for reps to implement than market segmentation because there are no judgment calls about whom to see next. Reps simply make the calls with the highest values.
In addition, the Call Value Targeting approach represents the most precise way to identify the most profitable field force size. When one knows how many profitable calls can be made, it becomes relatively easy to determine how many reps will be needed to make these profitable calls.
Call Value Targeting works only if the doctor-level models accurately forecast scripts. If the models don’t work, the resulting program will needlessly mess with the sales force and produce no improvements in the business. If the models are accurate and put to use, our analysis indicates that most companies could about double the amount of incremental business their sales force generates in a given year.
There is a quick and easy way to validate the model-building methodology for an in-line brand. First, give the model-builder 24 months or so of detailing and sampling data at the doctor level. Also give them the first 18 months of the corresponding doctor-level prescribing data. Then, tell them to forecast how many scripts each doctor wrote during the last 6 months for which the prescribing data were withheld.
In order to determine if the models accurately account for the impact of detailing and sampling, the doctor-level forecasts need a sort of placebo for comparison. To create the placebo, one simply assumes that each doctor’s prescribing during the previous period will remain unchanged into the period being forecast.
These placebo forecasts assume that detailing and sampling have no impact. If the model builder’s forecasts aren’t significantly closer to actual scripts than the placebo forecasts, the methodology that produced the models is bogus.
It is important to mention that simply correlating forecasted TRx’s with actual TRx’s can show a high correlation, even when the models that produced the forecasts fail to account for details and samples. This is because model builders will always take into account the volume of past script writing, and past script writing is correlated with future script writing, independent of the impact of detailing. Thus the need for placebo forecasts for comparison.
The ways of successful package goods marketers offer the pharmaceutical industry a lot. Above all, these companies strive to avoid relying on judgment. They try to keep from doing anything in the real world without first knowing what the result will be. Some pharmaceutical marketers relish judgment calls and waste a lot of money as a result.
Mass marketers spend a lot of time testing. They devote considerable effort to making sure they are testing the right things and using well-validated methods. One of the most respected companies took two years to experiment with different TV commercial testing methods before choosing a winner.
However, when it comes to marketing prescription drugs to doctors, market segmentation based on raw historical data is not cutting edge, even if it mimics the way P&G and Lever do things out of necessity. It is dumbing down the data.
S. Kent Stephan is the CEO at Princeton Brand Econometrics
SIDEBAR: What is Market Segmentation?
Market segmentation in-volves grouping individuals according to criteria that determines their placement into one or more categories that are believed to differentiate the potential to respond to promotion of a given brand. It is generally used to help allocate scarce promotional re-sources so as to achieve the greatest return on investment.
SIDEBAR #2: How Call Value Targeting Works
1 Builds accurate mathematical models to forecast how many incremental scripts each individual doctor will write in response to any number of details and samples over a defined period of time.
This is done for each promoted brand.
2 Validates the accuracy of the method used to build each brand’s models.
3 Integrates models for individual brands to find the most profitable sequence of brand details and amounts of samples on each potential sales call for each doctor.
4 Quantifies the expected value of each potential sales call in terms of sales or, better yet, gross profit.
5 Offers representatives instant access to this information to plan their activities with profit maximizing precision.