How to Calculate the ROI of Online Communities

The exact process you can use to calculate the return varies depending upon the metric you want to measure. However, most are based around a very simple methodology. We have outlined this methodology below.

Step One: Determine if the variable has changed over time.

First identify which of the benefits you’re trying to measure. This can be any of the metrics listed in the benefits above. Next, we try to measure this over a fixed period of time. This time frame might vary by benefit (retention rates might be by year, frequency of purchases, or average order value might be by quarter).

1. Use a sample of members. This might mean measuring the variable before and after members join a community. A typical interval here might be 3, 6, or 12 months. If members are spending $100 per year before they join the community and $150 per year after they join the community, that’s a measurable impact. We might track the spending habits of 100 members when they join the community, then track their spending of those same members 6 to 12 months later to see if the average spending has increased. We can analyze spending habits when they join either directly (i.e. looking at customer files) or via surveys (asking members to estimate their spending with the company in the previous 12 months and completing the same survey a year later). 

   If members spend $100 pa before they join the community and $150 pa after joining, that’s a measurable impact.

    

   Click to Tweet

2. Run an experiment. A more accurate method is to run an experiment. This usually involves segmenting customers at random into two groups. Half are invited to join the community, while half are not. This removes the self-selection bias (i.e. members who join a community are more likely to increase spending/buy more/be more loyal). Then we compare the differences in this behavior over a period of time.

This can be used for almost all the variables mentioned. These variables include retention, lead generation, spending levels, customer satisfaction, etc.

The critical step here is to first benchmark current (or past behavior) and then compare this with future (or now current) behavior to determine whether this has changed since joining the community. This will usually mean identifying the increase or decrease in behavior change (e.g. spending/retention rates/CSAT) per individual community member

Step Three: Convert And Generalize by the Total Group

The next step is to multiply this by the total group represented in the community sample. For example, if each member now spends an additional $30 that is attributable to the community, we can then multiply this by the number of members represented in our sample to understand the total return generated by the community.

This can be complicated if we’re using a survey technique. Those most likely to have increased their purchases are also most likely to respond to surveys from the brand. In many cases, only the most active members reply to a survey. This means the data only reflects the most active members. It’s very important, here, only to include the number of people represented. A common mistake is to multiply the increase by the total number of members. However, many members no longer participate and thus should not be included.

You could only use active members in the community (non-active members probably don’t matter as much). However, it’s important the survey reflects activity levels from all quotas of the community. If 1% are heavy users, 9% highly active users, and 90% simply lurk, then your survey data has to be comprised of the same group (or weighted to represent that group). Once you have this, you can multiply the average spending increase by the number of people in the representative group (usually the number of active members).

If you have 3000 active members, whose average spending has increased on average by $30, this is a total return of $90,000 per year. But return isn’t profit. The next step is to determine the level of profit.

Problems with this Method

Any method will incur numerous problems of accuracy. Any time frame for measuring impact is arbitrarily chosen. Asking members to estimate spending is likely to incur problems of accuracy. Spending is likely to change over time. This model also ignores members once they have passed beyond the time span. In practice, their spending habits may vary significantly. Thus, it would be ideal to track spending habits and comparisons on a cohort basis and total up the value of each instead of multiplying by a single metric.

This process will vary slightly depending upon the metric you’re using. The process is relatively the same. First, measure the change; second, determine how much of the change is attributable to the community; third, generalize across the broad sample; and, fourth, multiply by the average gross margin.

In the following breakdown of calculating the return from each mechanism, we have tried to highlight where this will be useful.