A new generation of analogues for valuing strategy
pharmafile | November 22, 2010 | Feature | Sales and Marketing | Alex Blyth, MSI, forecasting, marketing, marketing strategy
Anyone who is involved in the planning of pharma marketing campaigns will know that the Holy Grail is trying to interpret how a strategy on paper will translate into real sales. Robustly forecasting the returns of one strategy versus another is one of the most difficult tasks a marketer will undertake.
Consequently, management often backs strategies that ‘feel like’ they make sense, in the absence of numbers that show them what is sense. Sadly what makes sense is all too often, in the human condition, what we have seen done before and therefore feel comfortable with. The result is that ‘fresh thinking’, and potentially breakthrough strategies are seen with scepticism and exaggerated perceptions of risk.
Therefore potentially higher growth strategies do not get adopted, which is perhaps why so many plans in our industry look so ‘vanilla’!
So how can we change this and become a ‘Ben and Jerry’s’ industry, always coming up with fresh new recipes for success? The answer lies in being able to forecast doing things differently to demonstrate that the uncertainty and investment risk is in fact manageable. So how do we do this?
Many different models of forecasting have been developed over the years, but the same problem is associated with all of them: how do you validate those models, to turn mere crystal ball gazing into something which will more or less accurately predict the effect on sales of any given strategy? The challenge starts with the limited understanding of how ‘analogs’ – historical precedents – can be applied more inventively. Analogs can be a very useful way of validating forecast models and modelling what is typical. However, their application could be broader still if we knew more of the minutiae behind them, to value different strategies and Critical Success Factors (CSFs).
Essentially what is missing is the understanding of what was going on behind the actual sales numbers, and how much of the uplift we can attribute to each CSF.
The first part can be solved by interviewing the people previously involved with brands that are now being used as analogs, to understand the thinking at the time and the CSFs involved. By undertaking this commercial research first, you can start to build up a clearer idea of what was done when, from strategies to tactics, and impacting events or competitor activity at the time. Then you are in a position to understand what variables were at play – which information you can then enter into an econometric model to show what factor was responsible for what level of uplift – based on data correlations that back up and quantify what you have learned qualitatively.
I call these more sophisticated strategy analogs ‘Super Refined Analogs’. I believe they are the key to turning business changing strategy into robust forecasts that management will put their money on with confidence. Management just need to be shown ‘what strategy is likely to have the biggest impact on sales’, from launch right through to Loss of Exclusivity (LoE). If we are not doing this we cannot blame them for reverting to trusting their gut instinct regarding what works.
Start by ensuring the base forecast is solid
What I refer to as ‘the base forecast’ is for many where forecasting ends. It is the ‘typical forecast’ for a product in your situation doing the typical ‘vanilla’ strategy (e.g. the walk away at LoE we saw for simvastatin).
Before we can assess the returns of doing things differently we need to understand this.
There are four popular ways to build the base forecast:
1. You can go straight for using an analog, if for example your product is a ‘me-too’
2. You can use a situation based model built off a number of historical precedents essentially pushed through a regression model to make an algorithm where you enter your order to market, etc.
3. You work back from your peak sales potential and then choose a suitable uptake curve (e.g. S/D/C curves) – ideal for markets with poor market/historical data where you have done propensity to prescribe research
4. You build a patient funnel based model by understanding the journey and making informed judgements on what change is typical year on year – the ideal approach for new products or those with high change potential.
Regardless of how you build your base forecast, you will ultimately need a sales or market share analog to validate your base forecasts peak sales and uptake curve, as illustrated (above right). What you are hoping, if your forecasts are going to be accurate, is that the analog lines are more or less contiguous with your base forecast.
If there are big differences, then that raises a red flag about your forecast.
In this example our peak share analog (e.g. peak seen in similar sized therapy area) supports the base forecast, but our uptake analog (e.g. uptake seen another product with the same order to market) does not support the D-curve, instead suggesting a slower C-curve.
The need for a new analog
The big disadvantage of this approach is that it only tells you what is possible based on vanilla strategies which have typically been executed in the past. What it doesn’t tell you is the value of doing things differently.
Unfortunately it is common practice in our industry to look only at what will happen if we do the same-old, same-old, without considering the potential effects of new strategies. After all, how can you use analogs – essentially empirical snapshots of the past – to validate completely new strategies?
So you are left with a useful base forecast, from where you now need to work out what you might need to do differently, and what effect it might have. What is required is a framework which enables forecasters to understand what returns are possible from pursuing different strategies that give different levels of uplift.
This is when you need to add another layer into your forecasting model: the Super Refined Analog.
Perhaps the reason that the current best practice model doesn’t include this element is that it’s relatively difficult to achieve. The simple analog, where you just take things like percentage penetration, has the considerable advantage that the data is readily available from places such as IMS. It is just simple sales data, whether share of scrips or share of patients – it is easy to obtain. This is a good place to start, and certainly at the early stage of building a base forecast or a phase I forecast for a product, this is data that marketers should indeed be using. But by the time they get to phase IIb or accelerating growth in inline products, they really need to be thinking about alternative potential strategies.
The challenge with the Super Refined Analogs is that you need more data; not only that, but you need to have the background to the data, which means actually talking to, for example, product managers who have pursued different strategies with different products in the past. These analogs require a much more considered approach to data gathering.
The Super Refined Analogs
Super Refined Analogs require both the numbers of what happened, and the understanding of the qualitative insights to know why it happened. Because you are understanding the strategies behind the bald figures, they enable you to look at the returns of individual CSFs, as illustrated below. The base here reflects a slow erosion through generic entry, the CSFs shown reverse that trend by doing things differently.
So if we want to be able to understand the contribution each CSFs makes with confidence, how can we create these Super Refined Analogs? The key is talking to the people concerned with the historical analog product at the time, to understand the main variables which were acting on the market from their point of view. Then we can understand what they really did, and how much they really spent on what. We have found this a very rewarding task for all involved, it is after all an analog for somebody’s past ‘baby’ that often comes with some heroic tales of how to execute the strategy successfully, which the sales data alone would never tell you.
Armed with this qualitative data and insights, you enter these variables into regression analysis, knowing the stuff which really matters, which in turn enables you to understand what really drove the sales numbers.
However, if you are not an econometrics expert and are daunted by this, there is a simpler way – simply eye ball the data. Look at where a company started changing its strategy. Trend – or use a ruler – to look at where it was heading beforehand, and compare the difference the product managed to achieve above trend.
This is a good indicator of the percentage uplift that is possible from this strategy. It can be better still if you have two similar products that pursue different strategies consistently – you simply look at the divergence. It is not as ‘neat’, and the change can be due to other variables that a regression could have accounted for in the right hands.
However, most importantly the approach is a lot better than guessing – or worse still never bothering to even think about – what strategy or CSF can contribute what; which let’s face it is where most marketing plans are now. The result is that you can see the value of doing things differently, and focus your resources on those CSFs offering the best returns. The switch can bring in a new era of unleashed breakthrough growth strategies into the business. Marketers and business intelligence can do for the pharma business what Ben and Jerry’s creativity did for the ice cream business.
Real learning from the past
Tying together the strategic planning and the forecasting process remains one of the biggest challenges of pharma marketing. With so many potential variables having an effect on the success of different strategies, no model is perfect, and there will always be environmental factors which will skew the accuracy of any forecast.
But that is not a reason to avoid making sure that any forecast is based on as much good quality data as possible. The key thing is to know what difference each possible variable might make, and the weakness of pharma’s current approach is that it cannot allow for different possible strategies. We are choosing to walk with the light off, rather than reaching for the light switch at all. The excuse being that if you can’t see everything you might as well not bother, especially when you can just stick to the bits you’ve seen before!
At the moment, pharma marketers are good at picking the right analog to understand what is quantitatively possibly within any given market situation. What they are less good at is picking analogs to understand which strategic option is going to be most effective. In other words, they are good at doing analogs to find out if something is accurate; but less good at choosing the analogs to tell if a strategic decision is right. Currently we are a bit ‘copy & paste’ with data and rarely add much value beyond how IMS/CSD etc., simply prepare it. I see this as the equivalent of giving your child a lego kit to make a spaceship, and then only ever seeing them running around the room enthusiastically with the individual bricks, making lots of the right ‘whoosh’ noises – but not actually doing anything with it. By building strategy forecasts and using Super Refined Analogs, forecasters and business intelligence become true decision support scientists who will be worth their weight in gold.
I am keen to see this transition. There are only so many ‘whoosh’ lego bricks and raw data charts you can see in a presentation while managing to hold a smile.
Based on the increasingly talented pool of people I see in pharma, I think it is now time for real business intelligence.
Bring on the innovators.
Alex Blyth is a managing consultant and head of marketing sciences at The MSI Consultancy. Contact him at ablyth@msi.co.uk.
Related Content
Exscientia buys Austrian cancer cell screening company for €50 million
UK-based AI pioneers Exscientia has announced the purchase of rival AI firm Allcyte for €50 …
Deloitte acquires Iperion Life Sciences Consultancy
Deloitte have announced the acquisition of Iperion Life Sciences Consultancy, a globally-operating firm working to …
Generics could save US Medicare $1 billion per year
The United States national health insurance programme Medicare could have save $1 billion through the …
