Using key driver analysis with your survey data

Do you have great customer satisfaction ratings? Or is loyalty plummeting? Whether you’re celebrating your performance or working out how to fix it, the best way to assess it is to start by understanding what’s driving it. And that’s where key driver analysis comes in. Key driver analysis is a powerful technique that gives you an insight into the factors or drivers that are most important to customers and, therefore, have the greatest potential impact on your performance. 

Key driver analysis (KDA), which you may sometimes see described as relative importance analysis, essentially looks at a group of factors and weights their relative importance in predicting an outcome variable. It can be a big part of your market research. Outcome variables are usually performance indicators such as customer satisfaction, customer loyalty or Net Promoter Score (NPS). Whether these scores are high or low, it is useful to know the factors driving them. By comparing the relative importance of factors such as price, reliability or status, you’ll be able to answer questions such as:

Drivers are all of the factors that could potentially have an impact on your measure of performance. The factors that most meaningfully drive performance outcomes are known as key drivers. It is likely that you will have a combination of these, and the precise combination will differ according to your company type. For instance, if you run a tax preparation consultancy, the potential drivers affecting customer satisfaction might be:

What key driver analysis does is enable you to compare the relative contribution that each of these four drivers makes to the satisfaction of your customers. Each relative contribution is known as an importance weight and typically adds up to 100 (as in the example below) or to the R-square statistic.

In the example above, the ability of the tax preparer to increase tax refunds makes the greatest contribution to the satisfaction of the tax consultancy’s customer base. Of lesser importance is the professionalism of the tax agents. Insight like this can be crucial in helping you deliver great customer satisfaction or increase performance. 

Imagine, for instance, that as the owner of this business, you want to focus your efforts on improving customer satisfaction but you only have a small budget. Using the results of your key driver analysis, you might decide to focus on training tax professionals on the latest tax reforms and changes to deductions so that your customers stand the best chance of getting a refund.

But how do you get to this insight? First, you need to start with a survey.

Key driver analysis relies on survey data that captures some aspect of performance from the perspective of your customers. In the first instance, you’ll need to decide which output variable to measure. For customer-facing businesses, such as retailers and e-commerce sites, it usually makes sense to ask questions that capture customer satisfaction, e.g. “How satisfied were you with your purchase?” Other typical metrics are customer loyalty, repurchase intentions or willingness to recommend the business to a friend or acquaintance.

Next, you need to identify some probable drivers. These are also known as independent variables or predictor variables because they predict the main outcome. You should choose drivers that make sense based on the outcome metric you’re evaluating. Typical drivers are price, convenience, quality or packaging, or anything at all you expect to drive customer satisfaction, happiness or other measures of your performance.

Whatever you choose, it makes sense to use discrete variables to measure both predictor and outcome variables.  A discrete variable is a numeric variable that can take on a set number of values between two scores. For example, you may ask respondents to indicate their satisfaction with a recent purchase on a scale of 1 to 10.  Capturing scores in this way is vital when it comes to the correlation and linear regression analyses, which we’ll cover below.

Once you’ve gathered your survey data, you can start performing your analysis.

First, you should measure the weighted performance of each of the drivers from your survey. Exactly how each driver will be weighted will depend on how you have gathered your data. 

For example, let’s suppose you’ve asked your respondents how happy they are with their latest visit to your shop. Specifically, you ask them to rate the availability of products, the prices, the convenience of the shop's layout and the friendliness of the customer service on a 5-point scale from 1 (very unhappy) to 5 (very happy).  In this case, the weights for each driver will be assigned on a scale from 1 to 5 and the individual performance of each driver can be viewed as a percentage of the overall weighted score. If you asked 60 respondents these questions, you might end up with a summary of scores like this:

N=60

A correlation measures the existence of a relationship between two variables of interest: in this case, each predictor variable (independently) and the outcome variable. Each correlation is represented by a figure called the correlation coefficient, which can range from -1 to +1, with positive scores indicating a positive relationship between the pairs of variables and negative scores indicating a negative relationship. 

For example, if the correlation coefficient between the perceived level of friendliness of customer service agents and customer happiness is 0.15, this means that as customers’ perceptions of friendliness increase, customer happiness increases (or, as perceived friendliness falls, so does customer happiness). In contrast, a correlation coefficient of -0.15 means that customer happiness increases as friendliness decreases, which would be a strange, but perhaps not impossible, finding.

However, whether the value of the correlation coefficient is positive or negative is not enough to make inferences. The value itself is also important and measures the amount by which the overall metric will change for every one-point change in the driver’s weighted score.  In general, you can follow this rough rule of thumb:

Therefore, in our example above, a correlation coefficient of 0.15 between friendliness and customer happiness would indicate a low-strength, positive relationship. Now, let’s suppose we find that the correlation between availability of products and customer happiness is 0.8. This is a high-strength, positive relationship and would indicate that you should concentrate more effort on making sure your products are on the shelves at the right time than on making sure your sales agents are friendly.

The final step in key driver analysis is to use linear regression to determine the relative weight of each correlation between each key driver and the outcome variable being tested. 

Linear regression analysis works by testing all the pairwise correlations between the independent variables (the drivers) in order to yield the optimal linear combination that would predict the outcome variable. It distinguishes the relative contribution of each driver to the outcome and also generates an R-squared value which measures the contribution of all drivers together. The R-squared value can range from 0 to 1 and is converted into a percentage. The closer the number is to 1, the greater explanatory power the model has. For example, an R-Squared of 0.82 for our earlier customer happiness example means that 82% of the variance in customer happiness can be explained by our four variables (product availability, prices, shop layout convenience and friendliness). 

Another way to look at it is that 18% of the variance in customer service is unexplained and there are drivers (such as the length of the queue at checkout or the ease of parking) which have not been evaluated and which may also matter.

The results of your key driver analysis can also be visualised on a 2 x 2 matrix or chart. The y-axis shows your outcome measure (e.g. customer satisfaction) and the x-axis shows the driver’s degree of importance. Once you’ve plotted each driver against these two measures, you’ll find that they fall into one of four regions:

Drivers that fall in the upper-right quadrant of the matrix are the key drivers or critical attributes. These drivers play the most significant role in driving performance and are the area where you should be focusing investment and resources. 

The upper-left quadrant contains secondary drivers that are also important in driving performance but not as important as the key drivers. The lower-left quadrant contains low-impact drivers. Customers see these factors as unimportant, or these factors have a limited impact on satisfaction or whatever outcome you’re measuring. Finally, factors that fall into the lower-right quadrant of the matrix are the areas that are in need of improvement but less important to customers. 

As you can see, key driver analysis is relatively straightforward to perform, but it can be immensely powerful. This type of analysis has the following major benefits:

Sometimes, observing customers’ actions can lead you to believe that certain factors are important in explaining their habits and behaviours when, in fact, those factors are not very important. 

For example, let’s suppose you observe pub-goers visiting two different pubs on the same street. The car park of the traditional pub is full, whereas the car park of the gastropub, which you run, is quite empty. This might make you think that customers place greater weight on prices, making you wonder whether to drop your prices in line with those of your competitor. By conducting a survey-based key driver analysis, however, you might find that your pub’s car park is difficult to access or that your rival has friendlier members of staff. This is an actionable insight that can help you regain custom. 

As we’ve seen, key driver analysis not only tells you the bundle of drivers that affect your outcome of interest, but it also tells you the drivers that have the strongest influence. This is especially useful if resources and budgets are limited. For example, let’s suppose you have a limited marketing budget that could be spent on either revamping your food and drinks menu or a new set of advertisements. By identifying which of those would have the biggest impact on customers’ perceptions of your brand, you can focus your investment in a way that is more likely to improve brand performance.

Combined with customer attribute data, such as demographic data, you can compare the key drivers of your various customer segments. This is very useful if you expect different customer segments to have different preferences and drivers. For example, older pub-goers might place greater emphasis on traditional pub fare, whereas younger ones might be looking for a more extensive international offering of food and drink. Identifying the key drivers for each segment can help you target your messaging and marketing campaigns more effectively.

The data that our key driver analysis provides not only gives you an insight into the current key drivers, but it also actually forms the basis of a predictive model that can be used as a decision-making tool to run a number of what-if scenarios. For instance, you might use the tool to determine the impact on customer happiness if average prices were to fall by 10%.

Let’s take a look at some specific examples of where key driver analysis might be used. 

Satisfaction is a very common metric that is tested using key driver analysis. As we’ve discussed elsewhere, satisfaction is a major contributor to customer loyalty, and loyal customers are lucrative customers. Identifying the key drivers of customer satisfaction can therefore play a crucial role in driving sales and profits.

Another common outcome measure is the Net Promoter Score or NPS. The NPS helps you understand what customers really think of your company or offerings by posing one very simple question: How likely is it that you would recommend this company to a friend or a colleague?

Answers are usually recorded on a discrete range from 0 (very unlikely) to 10 (very likely). What’s great about the Net Promoter Score is that it helps you understand whether customers are willing to advocate for you, and this is very cheap but very effective marketing! Learning the key drivers of NPS can therefore help you increase your customer reach and build your brand. Read more about how to calculate your Net Promoter Score.

Purchase intent measures whether customers have an intention to buy your product in the near future. Applying a key driver analysis can help you determine the factors that shape buying behaviour, which will help you identify the factors that should be manipulated to push sales. You can assess the attributes of your products or services as well as the attributes of your competitors. For instance, you might ask about your prices and the prices of a near-substitute, using KDA to determine the relative influence of each on purchasing habits. This is vital information to support your competitive strategy. Purchase intentions can also be captured from prospective rather than existing customers. 

Sometimes, sales of products drop dramatically, which might leave you scratching your head as to why. Key driver analysis can play a key role here. It will help you understand and compare the drivers of multiple categories so that you can better understand where to focus your efforts to bolster sales. To do this, you might administer a survey that asks the same questions about different product categories. For instance, if you have two laundry detergents (a biological and a non-biological version), you can compare the relative impact of the prices of each on the performance of each brand.

When trying to improve your performance, it can be challenging to know what to prioritise and where to focus your investment. As we’ve seen, key driver analysis helps you understand which factors really, really matter to your customers and compare the performance of each factor. From there, you can make much better-informed decisions about where improvements need to be made or what needs to simply be sustained. Ready to start your survey? Our marketing solutions will help you out. Get started now.