Creating and Using Reference Groups

One of the most critical steps to successfully benchmarking an industry is having validated reference groups. Once all the required data has been collected from member businesses, and their data validated, improved and distilled it is now time to create an overview or picture of the results.

Now, this is a critical step for two reasons:

  1. It allows you to understand the relationship between all the various data points.
  2. You get to see the industry from a so-called bird’s-eye view. Which if anything should give insights into where the industry is headed and whether or not it is growing. Obviously, it isn’t breaking new ground for most associations that mostly track industry trends, but it still happens to be a very important step.

In this article, we will look at how exactly you can go about creating reference groups most efficiently.

Creating Reference Group Classes and Associated Reference Groups

The first thing you will need to do is define which of the reference group classes will be included in the benchmark. The classes will consist of member segments, so it needs to be well thought of before execution. You will also want to consider the size of each business, i.e., the number of employees and the revenue they are generating. Also, worth considering is the size, function, age, department, strategy, and cost, etc. So, every class will need to be a part of a question in the original questionnaire.

Once the above data has been highlighted the next step is defining the reference groups. The meaning of a reference group is a bunch of participants who have answered the question underlying the reference group class similarly. In a way, they have categorized themselves in a particular group by answering that question. That group will be a reference group. Defining the reference group or groups will require that you take the culminated data on a question underlying the reference group class and sort them into different groups.

Tip: You will need to import all the data from your benchmarking software into a spreadsheet. Some benchmarking software have an export feature, but others don’t. Try using software that supports this feature, or you’ll need to export the data manually.

Sorting Though the Data to Create Reference Groups

Now that you are aware of how to create reference groups, any method that you find easy can be used to sort through the data. However, the most effective way in our opinion is to use a spreadsheet program like Excel. Then sort the data of the underlying question(s) based on the reference group class you want to create.

For instance, you want to create a reference group class which is based on the size of a business. That size depends on the question ‘number of employees’. So, you can use Excel to sort the answers for ‘ the number of employees’ in descending order. You can then take the top 25% and sort them into the largest organisations, take the next 25% and sort them into medium-sized organisations, so on and so forth.

The one thing you will want to keep in mind is that each reference group should have enough participants to define the group. The participants should be able to contribute to the group. So, having just two or three participants isn’t going to be enough. Each group needs to have at least ten if not more participants.

If tiny groups are created, then they can be combined or discarded. If the groups are too large, they can better be split

Tip: If you’re in an unusually small industry where creating large enough reference groups is difficult, perhaps create one or two groups at most. It will still be useful. Though in some sectors not having enough data can make creating reference groups impossible in which case this can step can be foregone.

If you use a sophisticated benchmarking survey tool then you can create the reference groups within the tool. Based on the conditions per question the participants will be assigned to the correct reference groups automatically.

Group Results for Each Reference Group

The next step is to define the group results for each group. Again, a spreadsheet program like Excel happens to be the most effective, unless you use a sophisticated benchmarking survey tool. The group results for each group can be downloaded to Excel. You then either calculate or include the result per reference group for every variable by considering all the essential stats like mean, median, minimum, maximum, worst, best, number of participants, etc.

Now for every variable used you will want to know if the highest or the lowest is the best. The grade (highest or lowest) impact of the variable on the participant’s position. For instance, if you have a participating business which performs above 75% (the third quartile) for a positive variable, like revenue per employee, it is regarded as good. A participant which is performing above 75% for any negative variable whereby being the lowest is best, like cost per employee, is either a poor or very poor performing participant compared to the reference group.

If you have managed to conduct the benchmarks earlier, it would be helpful to include group results of all previous periods. The periodic information will allow for a comparison of the period results for every variable.


Defining, creating and then using reference groups is a fundamental element of successful benchmarking. Without this one very important step it can be hard to benchmark businesses in a broader more diverse industry.

For instance, benchmarking businesses in the retail industry will undoubtedly help highlight which segments of the industry are doing better and which are the worst performing. Though arguably it may not be as effective with smaller niche industries like retailers only selling designer shoes. Also, it isn’t going to be useful where there aren’t as many participants who define a large enough group. So, all of this will need to be considered when creating reference groups.

Learn more

Other articles you might like

How to source data for benchmarking

How to Source Data for Benchmarking? Once you know what to benchmark the next step is actually to start the benchmarking process...

How Benchmarking Can Help Business Associations Make Money and Attract Member

It is interesting to note that many of the reasons why businesses join an association correlate with reasons to benchmark...

Why Some Benchmarking Efforts by Associations Fail

Failure is often considered the route to ultimate success. However, associations don’t have that luxury...

How we support you

We believe that benchmarking could be much simpler, cheaper and more efficient through professional automation. And we believe you can add much more value for your clients with intelligent automation.

  • More than 15 years of experience in benchmarking
  • Help on implementing the questionnaire
  • Support on analysing and reporting the results
  • Training you to conduct your own benchmarking surveys successfully