Please login to the form below

Not currently logged in
Email:
Password:

Get more out

Pharma companies must learn to extract more value from existing products

On average pharma companies spend 30 per cent of turnover on promotion and this figure has not changed appreciably in 20 years. This is difficult to understand given the determined effort, over the same period, to improve the productivity of R&D, which still only accounts for a mere 15 to 16 percent of turnover.

Now, things are changing and management's attention is increasingly focused on this 30 per cent. This marked shift in emphasis accounts for the plethora of salesforce effectiveness managers employed in the industry today. The reason is clear. Despite combinatorial chemistry, high-throughput screening, dual tracking, genomics and a host of other clever interventions, R&D is failing to produce the quantity and quality of NCE's necessary to sustain the industries current size and to meet shareholders' expectations.

Pharma's new mantra is take cost out of the system and increase productivity.  To survive the industry must learn how to extract more value from existing products. To achieve this means improving the productivity of sales and marketing investments. Recent research suggests that this is feasible, as estimates show that between 30 and 50 per cent of money spent on promotion is wasted. This means that if an average company can improve productivity by 30 per cent - the lower of these estimates - then nine per cent will drop straight to the bottom line.

Unfortunately these returns are not happening in practice. There are two main reasons. First, a lack of clear unambiguous objectives, with several companies struggling to operate under different definitions of ROI. Second, companies are simply using the wrong kind of data analysis.

The simplest and most useful definition of ROI is Ä out/Ä in - a simple ratio of sales accounted for by that promotional vehicle, divided by the cost of the promotional activity. Forget patient-centred metrics ñ they are confusing and difficult to apply in practice.

What is the question?
A related problem is uncertainty over the questions you actually want answers to. Are you interested in understanding the overall effect of a given activity over a set time period? Or do you need to know the current incremental effect on sales?

The first is relatively straightforward the latter requires a much more sophisticated form of analysis. To use ROI as a strategic tool to gain sustainable competitive advantage you need to appreciate both perspectives.

As a minimum, companies should be able to rank their promotional investments in order of their effect and know which combinations actually produce the strongest synergies. This allows you to practice zero budgeting and forget about just using the previous year's budget template as a blueprint for this year.

A number of benefits accrue from this information alone. Promotional investment can be directed to where it will produce the greatest effect, which means that you will maximise sales and keep wastage to a minimum. You can easily build a promotional model which will allow you to test assumptions and clearly demonstrate how your promotional plan fits together. Then if corporate priorities dictate a cut in promotional spend, these can be made where they will do the least harm.

Not adding up
The second reason for ROI results not producing the desired benefits is simply that most companies use the wrong method to calculate ROI. Despite the fact that linear regression was invented before Bayer first patented aspirin, this remains the most common choice for ROI analysis.

Linear regression is fine for calculating weights of sweet pea seeds, as used by Francis Galton in 1875. It is not however suitable for ROI analyses given the complexity of modern pharmaceutical data. This is because linear regression relies upon a number of assumptions - the violation of any of these can render the results invalid, inaccurate and, effectively, meaningless.

The first assumption is that relationships, such as that between sales and calls, are linear. This is rarely, if ever, the case in my experience. Normally pharmaceutical response curves follow an S-shaped curve, an inverted U relationship or a V-max type curve - where the upward curve ends in a plateau effect.

The functional relationship can be coaxed towards a straight line, to some extent, by transformation using logs or square roots but this effectively removes the 'tails' from the regression line and forces interpretation on a few points in the centre of the data distribution. The tails at each end of the graph often describe the most interesting and most recent part of the relationship. Also, transforming the data in this way may render the results difficult to interpret in practice.

Erroneous assumptions
It is often assumed that the variance (the spread of the points or individual observations either side of the line) is equal. Unfortunately this is almost unknown with pharmaceutical data, where larger values tend to have a larger spread. If the variance changes along the regression line this renders the usual standard errors, making tests of significance and confidence limits untrustworthy and misleading. This is very common and one key reason why most companies believe that their promotional instruments are working better than they actually are.

Another common misapprehension is that each observation is independent and not correlated to any of our other points. For pharmaceutical data this is not true. Sales managers rely upon high call frequencies to drive sales, so this month's calls will be related to last month's calls. Similarly, to use linear regression we are assuming that each month's advertising effect stops in that month and that there is no carry-over. Clearly this is nonsense. More importantly significance tests all rely on independence between points which explains why linear regression consistently overstates the significance of our promotional effects.

Linear regression relies on the assumption that our data is drawn from a normal distribution and that our residuals (the difference between our regression line and the actual point that it describes) are similarly distributed. This is rare with pharmaceutical data and consequently  significance tests - such as t and F tests - which assume data is normally distributed, will return invalid results.

Even one larger than average point on the graph can considerably pull our regression line up, skewing the results.  Linear regression is not resistant to outlying points, perhaps caused by the exceptional rep whose call rate or meeting rate is much higher than average or by an aberrant sales figure.

Omitted variables are potentially even more damaging, especially if you are trying to draw causal relationships like 'higher calls on target doctors lead to higher sales.' Such variables may include doctors who have attended a conference and listened to an influential speaker. These can lead to overstating or understating of  the true relationship between target calls and sales.

Update the data
ROI analyses have not exerted the desired effect on our sales and marketing productivity because we have failed to ask the right question or we have used inappropriate, antiquated analysis. If you are using any form of ROI analysis based on an Excel spreadsheet you can be fairly certain that linear regression underpins it. This will render your results suspect, which perhaps explains why changes made in the light of your data fail to achieve the expected result. Further issues arise when you try to implement information based on country-level analysis in a regional setting, where the underlying healthcare environment may be very different.

The pharma industry prides itself on use of knowledge but when it comes to ROI analysis the bulk of companies still rely on analytical methods that were in use before our industry was born. Turning data into effective, actionable information is the key to sustainable competitive advantage. Modern computing technology coupled to more sophisticated analysis can provide the means to reduce sales and marketing costs by 30 per cent and add that nine per cent to your bottom line.

The Author
Dr Graham Leask
is a member of the Economics and Strategy group at Aston University - g.leask@aston.ac.uk

30th May 2008

Share

Featured jobs

Subscribe to our email news alerts

PMHub

Add my company
M3 (EU)

M3 was founded in 2000 with the goal of changing the world of medicine through making full use of the...

Latest intelligence

PM Society Digital Awards – the power of together
Our chief executive, Emma Statham, writes about the value of awards and the power of together....
Seduction_feature_image_thumb.jpg
Seduce anyone in four simple steps
You know the health of the global economy is dependent on our ability to seduce one another – don’t you? And you know that we need to be able to...
What Would Jeremy Do? : Assessing the impact of a Corbyn-led Labour government
GK Strategy are delighted to announce the launch our latest briefing paper entitled ‘What Would Jeremy Do? Assessing the impact of a Corbyn-led Labour government’....

Infographics