Two basic approaches are used to answer this question – stated importance and derived impact. The stated importance method lets customers rate what is important to them. The derived impact method determines the influence customers' ratings of various items have on an outcome of interest (e.g. loyalty, satisfaction) via statistical analysis like regression or correlation.

The derived impact method is generally the method of choice. It has a number of attractive characteristics in comparison to stated importance. It requires fewer questions (performance and impact information can be obtained from the same question), and it provides a predictive framework for determining how performance changes can impact customer perceptions. The most significant benefit of the derived impact method is a more discerning view of how factors impact customers' perceptions and behaviors. Customers will often state that certain factors are important, but they end up having very little impact on customers' perceptions and decisions.

However, the derived impact method has two main weaknesses. First, like all statistical techniques, it requires a minimum number of responses to produce robust results and the results are at an aggregate level. Unlike stated importance, impact scores cannot be calculated at the individual customer level. Derived impact scores indicate which factors are most influential to the average customer but do not tell us what influences each individual customer.

Second, the derived impact method provides an estimate of the influence an item has on something else, not the importance of an item to customers. This weakness is often overlooked. In fact, most researchers and practitioners refer to the method as derived importance instead of derived impact. This may seem like a small point, but it is important to understand. Let's say order delivery does not have a significant relationship with repurchase intentions. Does this mean order delivery is not important to your customers? Absolutely not. It means order delivery is not a key decision factor for most customers. If you have a major failure in your delivery process next month, order delivery will likely have an impact on repurchase intentions in the next survey analysis.

The two weaknesses of derived impact method are areas of strength for the stated importance method. With stated importance we can determine what is important to each individual customer while also aggregating responses to determine what is most important to all, or a segment, of customers. So the stated importance method may have a place in the savvy market researcher's tool kit after all.

But how do we overcome the main weakness of stated importance methods – customers tend to rate everything as important? There are a number of techniques available to minimize this problem. Factoring in concerns like between-item discrimination, number of questions, time to complete, and ease of analysis, I have identified the top three techniques for asking stated importance questions – Q-sort, maximum difference scaling (MaxDiff), and point (chip) allocation. If you want more information on each of these techniques, including detailed advantages and disadvantages, please click here.

MaxDiff is quickly becoming the preferred method of stated importance for many market researchers, but there is no "best" method for collecting stated importance in all contexts. Be sure to understand the needs and weigh all of the advantages and disadvantages of various techniques before choosing one.

How do these new stated importance techniques affect the age-old debate of stated versus derived methods? There really shouldn't be a debate. Stated importance and derived impact actually provide answers to slightly different, but complementary, questions. Stated importance tells us what customers think is important while derived impact tells us what influences customers' perceptions, intentions, and/or behaviors.

In reality, these two methods can work together to provide us with a better understanding of our customers. The figure below indicates what we can learn from integrating stated importance ratings and derived impact scores.

Since the upper right (Key Drivers) and lower left (Low Yield) quadrants agree (high/high and low/low) between both approaches, either approach could provide this information. The truly valuable differences lie with the other two quadrants.

The lower right quadrant (Essential Requirements) shows minimum requirements of customers or areas of competitive equity. These items are similar to airline safety, restaurant cleanliness, or cars with wheels – they are required before customers even consider your offerings but do not factor heavily into customers' decision making processes or overall perceptions. Items in this quadrant could be either potential differentiators or potential areas of overinvestment depending on whether or not there is demand for better performance or additional offerings.

The items in the upper left quadrant (Value Added Priorities) have high derived impact and low stated importance. These are primary opportunities for your organization. Most likely, all companies are focusing on the high importance items – they are the items customers are always talking about. The low importance/high impact items could be overlooked if you are not paying attention, but they are key differentiators and influencers for your customers. They are the items that delight your customers, the things they may not need or ask for but end up differentiating you from competitors. Focusing efforts on improving these items can often yield a greater payoff than efforts in other areas.

The derived impact method is definitely the best method for determining what drives your customers' perceptions and behaviors, but we should not discount the potential value provided by stated importance methods. Gathering stated importance data can certainly provide additional value for understanding your customers if it is used the right way. Like any tool, stated importance and derived impact can do more harm than good if used the wrong way or to answer the wrong question.