In May, the quantitative team at BHI reminded us of the importance of giving due thought and consideration to what you want to get out of your quantitative research as this ultimately affects what you put into the research (i.e. the questions). Then our focus was on the Max Diff Scaling Technique, this month the quant team review Key Driver Analysis.
Vicky McLellan 
15 Jul, 2008
Quant Corner: Key Driver Analysis
In
an increasingly competitive and cost focused market, one of the most frequently
asked questions by our clients is what drives our customers and how does our
brand perform on these parameters. Answering these questions effectively enables
marketeers to focus on the factors that will make the most difference to their
brands performance whilst also enabling marketeers to maximise the potential
of their resources (which can some be somewhat limited at times because of
budget constraints) by focusing on what really matters to their customers.
One of the most common ways of assessing what is of most influence to HCPs is to assess a number of attributes on a rating scale (BHI recommend a 7 point scale). This exercise is then repeated for the client brand and its key competitors to assess how well each brand performs on the most important parameters. Typically the analysis will focus on looking at the mean score for each of these attributes to map out where performance is high and where it is low.
However how well an attribute performs may not bear any relation to how much its performance contributes to the overall perception of a brand. Indeed physicians may have a very positive perception of a brand’s sale representatives but they may not rate a drug according to its representatives. Also if there is little differentiation in a market place a mean score in itself can become somewhat meaningless.
Inferring whether the performance of specific brand attributes impact on the overall brand performance is crucial for marketing decisions, and this is exactly what a key driver analysis aims to do.
Below is the most common output of a Key Driver Analysis.
So how does a key driver analysis work?
A key driver analysis is first of all a statistical technique. It usually involves what statisticians call a correlation analysis and/ or a multiple regression analysis. While you don’t need to be able to calculate these statistical analyses, it is useful to basically understand what they refer to in order to interpret the results.
A correlation analysis measures how strongly one measured attribute is associated with the overall performance of a brand. The general rule of thumb is the higher the correlation the stronger the relationship between the attribute and the brand. The attributes with the highest correlations are your key drivers.
Although a correlation analysis is easy to execute the main drawback is that it only takes into consideration one attribute at a time when comparing it to the brands performance. This can be a problem when it is not just a single factor that influences a particular correlation and as we are all aware decisions such as will I prescribe this drug tend to be based on a number of different factors (both emotional and behavioural).
In this case, a multiple regression analysis may help: as it takes into account separate attributes in order to explain a brand performance. Statisticians’ pet peeve here will be to have drivers that mutually influence each other... which usually happens when surveying about pharmaceutical products (the dosing of a drug may affect its side effect profile and vice versa). At a slightly higher cost, what is called a factor analysis may then be necessary: this will gather inter-correlated attributes into a new group, now independent to all reminding attributes. This way, ensuring your key driver analysis is most reliable.
Top tips to consider when conducting a key driver analysis
- Think carefully about which attributes you want to test before conducting your research. A key driver analysis works most well when testing a smaller number of attributes therefore it may be an idea to consider a round of qualitative research to ‘pilot’ the attributes amongst your customers to prioritise the attributes of most importance
- Ideally 5 to 10 questions should be asked in relation to each attribute (hence a key driver analysis may not be the best option when testing too many attributes). Questions typically focus on the use of rating scales.
- Remember that as with all statistical studies, large and representative sample sizes are often key to reliable results.
- Bear in mind when budgeting for your study that a key driver analysis will be an additional cost which will be dependent on the exact requirements of the study
- Seek advice from your statistician/ market researcher about your different options and results. Hence choose a statistician/ market researcher who is a good communicator as it is crucial to ask the right questions in the right way
Your response:
Login to respond to this article: