How To Use impact / Derived Importance

Impact, or derived importance, tells you which items on your questionnaire are most strongly linked to overall satisfaction (or likelihood to recommend, if you use NPS).

This guide gives you everything you need to know about using derived importance, what it tells you and what to look out for, and how to calculate it.

You can download a PDF copy of the guide here.

Download PDF Copy

What is it?

It is based on the correlation between the satisfaction score for each item and overall satisfaction, and is expressed as a correlation coefficient between 0 and 1. Think of each correlation coefficient as a summary of a scatter plot showing the relationship between that item and overall satisfaction:

When the correlation is high people tend to score overall satisfaction high if they score the driver high, and they tend to score overall satisfaction low if they score the driver low (as in the example on the left). When it’s more moderate, you’ll find that customers do not necessarily score the driver and overall satisfaction in a similar way.

What it TELLS YOU

The higher the correlation, the stronger the link between that satisfaction score and overall satisfaction.

If we assume that overall satisfaction is a result of the sum total of your performance on each individual aspect of satisfaction, then we can say that the items with the strongest impact are having more influence on the overall score.

When customers think about how satisfied they feel, those are the things which are weighted most heavily in their decision. In other words, those are the key drivers of overall satisfaction, and they are what makes the most difference to how satisfied customers are with the overall experience.

Another way of saying this is that if we can change how customers feel about those key drivers then that’s what will make the most difference to overall satisfaction.

Friendliness of staff, in this example, is the strongest driver of overall satisfaction. That doesn’t mean it’s more important than product quality, but it provides more differentiation right now.

HOW TO CALCULATE IT

There are more sophisticated measures (you can find out about some of the options in our free webinar Best Practice Driver Analysis), but the best starting point is a straightforward Pearson correlation.

=CORREL()

Excel Function

You can use the '=Correl()' function in Excel, if you don’t have access to statistical software. For each impact score you’ll need to correlate your entire range of satisfaction scores for that driver with the matching range of scores for overall satisfaction.

The formula expects two matching arrays of data, as shown in this example. If your data is arranged in columns, with each customer’s responses on a row, then the first part of the formula will point to the column of scores for the driver (B2:B201 in the example below with 200 cases), and the second part will point to the column of scores for overall satisfaction (C2:C201). These two arrays should always be the same size.

WHAT TO WATCH OUT FOR

Impact is a simple measure to calculate, but it needs to be interpreted with care or it can lead you into damaging mistakes. Here are a few things to watch out for...

IMPACT IS NOT IMPORTANCE

We prefer the term “impact” to “derived importance” because this technique does not measure what’s most important to customers, it measures what’s making the most difference right now.

Items which are very important may have low impact if your performance is consistently good. If you take your eye off the ball because of that low impact, and performance drops, then you’ll see impact rising rapidly and immediate damage to customer satisfaction and loyalty.

Impact is a good way to allocate extra effort, but not a good way to reallocate effort.

IMPACT IS VOLATILE

To pick up on the previous point, you will see impact changing over time, much quicker than stated importance scores do. If you experience sudden supply chain problems, for example, you can expect to see the impact of items such as availability and lead times becoming stronger overnight.

IMPACT ONLY CONSIDERS ONE DRIVER AT A TIME

The impact correlation looks at each driver in isolation, asking how strongly it associates with overall satisfaction.

If your questionnaire includes items that are very similar (for example, friendliness of staff and helpfulness of staff) then you are effectively double counting their impact on overall satisfaction.

More sophisticated techniques (e.g. relative importance analysis) take this into account, but if you rely on correlation you need to make sure your questionnaire is as lean as possible, and that you allow for overlap between similar items when you interpret the drivers.

YOU NEED ENOUGH SAMPLE FOR A RELIABLE MEASURE

Like everything else on your survey, correlation coefficients come with a margin of error. You need a sample size of at least 50 in each group to have a robust understanding of impact.

Summary

Overall, impact is an easy and useful measure that can help to understand what’s going in customers’ minds when they think about their overall relationship with you. Used alongside stated importance it helps you to prioritise based on what makes a difference to customers, without losing sight of the importance of the givens.

+ Pros

  • Easy to understand & communicate

  • Robust & sensitive

  • Reflects customer needs

- Cons

  • Dependent on an accurate list of attributes

  • Requires importance and satisfaction scores

  • Can be complex to calculate with missing values

Want to Know More?

Send us a message if you would like to learn more about analysis techniques in customer research.