How to Calculate the ROI of Online Communities

Many communities created by passionate fans and hobbyists aren’t done so to create or keep customers at all. The majority of communities in the world were created by passionate amateurs. Their goal (when one exists other than enjoyment of the subject) is to generate advertising revenue. This also applies to many communities on major news sites (such as those that allow commenting).

A community can increase ad revenue in two forms. This might be advertising driven by repeat visits as a result of community features (e.g. commenting) added to the site, or advertising revenue via placement adverts on the community. This latter figure can be taken directly from advertising invoices. The former requires a calculation to identify the advertising revenue per single page view and then the number of additional page views generated as a result of adding community features.

To calculate this, you need the following data:

1. Percentage of repeat visitors on a random selection of articles with comments enabled.
2. Percentage of repeat visitors on a random selection of articles with comments disabled (this is explained below).
3. The total number of visits to comment-enabled articles per month.
4. The CPM rate (cost per thousand). If using CPC (cost per click) or CPA (cost per action), this will be very different.

Step Two: Measure the Difference in Repeat Visits

Here, we need the increase in repeat visits on articles with comments compared with the number of visits to articles. This can be achieved in three ways:

1. Split test between visitors. Hide comments from some members and measure the difference in return visits between the two. This can then be multiplied by the total number of visits to calculate a rough idea of how many additional visits a comment section calculates. In practice, the ability to hide comments from a random sample of members is likely to be available to a limited number of people looking to calculate the benefits of comments.
2. Split test between articles. This would mean disabling comments on a random selection of future articles and comparing the % of return visits between two comparable samples. In practice, this is typically the easier of the two methods to use. If, for example, the % of return visits to articles without comments is 46% and the % of return visits to articles with comments is 49%, we might attribute the 3% difference to the comments. We can now state that, on average, every article with comments enabled causes an additional 3% in additional visits through return visits.

For our purposes, we are using the second method. By calculating the average between them, we can calculate what % of visitors return to the community because comments are enabled.