How to Calculate the ROI of Online Communities

By Richard Millington

ROI People

By far the most common reason to launch a branded community today is to reduce customer service costs. This has a clear and direct link to revenue. If your customers can find the answer to their question in the community before contacting a paid staff member, you can theoretically reduce the number of customer service staff members (or hopefully find them cushy, yet important, jobs elsewhere in the business).

This is usually measured by direct or indirect call deflection. We’re going to make an assumption here that call deflection or indirect call deflection also includes any contact with the customer service agent. This can include email, instant chat, showing up at the office and shouting angrily, etc.

To measure this, you need a lot of data. This includes:

  1. Number of questions answered in the community.
  2. Percentage of questions that were answered satisfactorily.
  3. Percentage of satisfactory answers from users who would otherwise have called customer service.
  4. Possibly percentage of customers in warranty.
  5. Number of users to resolved questions pages in the community.
  6. Percentage of these users who gave satisfactory answers who would otherwise have called customer service.
  7. Possibly percentage of this group in warranty.
  8. Total customer service staff costs.
  9. Technology costs associated with customer service.
  10. Allocation of fixed overheads.
  11. Total number of calls (all contacts) received per year.

In practice, you might want to use the same answers for 6) and 7) as you did for 3) and 4), as the answers are unlikely to vary much.

Community ROI Template

You can drop these figures into this spreadsheet here or follow the process below.

Step One: Measuring the number of calls deflected

Calls can be deflected in one of two ways. The first is someone who can ask a question and receive a response via another member in the community. This is known as direct call deflection. The second is someone who can see an answer in the community and, therefore, doesn’t need to ask the question. This is known as indirect call deflection. We need to combine both of these to measure the number of calls deflected.

Direct call deflection can be measured in one of three ways:

  1. Run a withholding test. This involves either comparing the call volume before and after the community, or closing the community for a specific amount of time and measuring the number of calls received during this period.
  2. Tracking the number of questions answered. One method is to measure the number of questions that are answered by other members of the community. This is usually the easiest metric to track. However, it fails to consider whether the problem was resolved. For example, someone might get an answer in the community yet still not solve the problem. They might, therefore, then contact the customer service line.
  3. Tracking the number of questions resolved. The best method is to measure the number of questions that have been resolved. This can either be a feature offered in enterprise platforms or via a pop-up survey asking members to highlight if a) the problem was resolved and b) if they otherwise would have contacted the customer service line. We can then multiply the number of questions answered by the % who received a satisfactory answer by the % who would otherwise have contacted customer service.

Note, some organizations also multiply this by the % of customers still in warranty. This is not a distinction we make in our calculation.

This figure will give us the total number of direct calls deflected.


Indirect Call Deflection

In addition to the direct call deflection, we also want to measure indirect call deflection. This is the number of people who viewed questions that were answered by others in the community. This can be measured by tracking the number of visitors (users) to resolved questions in the community and then following the same process.

Notice here that it is important that surveys or pop-up quizzes exclude the original poster of the question so as to avoid double-counting (or surveys that ask people the number of questions asked, satisfaction rate, and whether they would call the customer service line can be used as a backup if on-site surveys are not possible). If we can’t do this for all visitors to these pages, we may instead use a sample of pages.


Step Two: Calculate The Total Number Of Calls Deflected Per Year

Once we have the metrics above, it is easy to combine the total number of calls together to determine the total number of calls deflected as a result of the community per year.


We can see, for example, in the first year the community directly deflected 16,625 calls. Now we need to find out how much each call is worth.

Step Three: Determine The Value of A Call Deflected

Once we know the number of calls deflected, we need to assign a value to these calls. This can be calculated by dividing the total cost of the current customer service call center (people + technology + overheads) by the total number of calls received during this period.


We have assumed that these costs won’t change. In reality, they will. However, this saves us time by calculating for each individual year and provides a proxy answer we can use.

Step Four: Determine the Total Value of Calls Deflected

Once we know the total numbers of calls deflected and the total cost per call, it is relatively easy to calculate the total cost saving through calls deflected by multiplying the two, as shown below.



  1. Call deflection is the most common reason to create an online community and the most widely used method to calculate the value of a community.
  2. To measure call deflection, you need to determine the total number of calls deflected. This isn’t just the number of questions answered or answers viewed, but also the % of these whose problem was satisfactorily answered and would otherwise have called the customer service line.
  3. You can also multiply this answer by the number whose products were still in warranty. However, we advise against this.
  4. Calculate the current cost per call and multiply this by the number of calls deflected to get the return generated.



 LinkedIn LinkedIn

 Google Plus Google+