Using analytics to understand visitor retention

I wanted to spend some time today talking about visitor retention (aka Cohort Analysis) very briefly.

Our cohort analysis graphs are fairly simple, the y-axis shows what percentage of users come back and respond to an event while the x-axis is the data for a cohort. So for example with the “Bad retention” graph, all users that came to the site on the 31st, only ~7.3% showed up the next day, and ~5.7% showed up 2 days later, so on and so forth.

Bad retention:

This means users come to the site and rarely ever come back possibly telling you, it’s probably not that interesting or very valuable to use after the first day of use. Super viral applications generally show trends like this because they have great flow and viral mechanisms but they do not last.

Decent retention:

While it could be better by getting a higher percentage as the highest is between 25-35%, this site shows promise as the cohorts (each line) fish tails upwards connoting some stickiness to the site or reason to come back even as much as 8 days later. Providing more value to the product could help the cohorts in the 2-5 day positions.

Good retention:

While it’s not the epitome of good retention, the graph says a lot about what is going on. The site appears to be sticky for 3-4 days and brings nearly half its visiting users back sometimes. Additionally, some of the cohorts are actually flat and retain the same percentage the following days which also fairly impressive as most cohorts would show a decline eventually.

The best would be if it fish tailed like the previous graph we talked about above.

Mixpanel offers retention analysis like this for any features of your site not just if a user comes back. For example, we know what % of people come back to look at our demos and we think this is one of the most valuable tools we can offer for business insight for startups and companies alike.