This post was originally written by Tim Trefren on
Because a range of our customers are social game developers, we can get a
high-level look at trends theyÂ’re seeing in their Facebook applications. One
of the big trends weÂ’re seeing is that games are using tutorials to generate
strong retention among new users. A related trend is that this initial
retention is critical to the health of your game, in the weeks following
launch. HereÂ’s a closer look.
Impressive Results From Tutorials
One thing weÂ’re seeing succeed is the
tutorial-based signup process. A well-crafted tutorial removes all the
ambiguity out of getting started and helps teach a new user how to play the
If youÂ’re not familiar with this technique, the FarmVille signup process is a
good example. FarmVille explicitly teaches you how to harvest, plow, and plant
seeds with a 3-step tutorial.
Now that youÂ’re familiar with the concept, letÂ’s take a look at the data IÂ’ve
compiled from a number of games.
By The Numbers
The most impressive finding of this analysis is that individual steps in a
tutorial convert at over 90% on average. Meaning, once a user has started a
tutorial, they have a greater than 90% chance of continuing at each step.
This doesnÂ’t include the first step, however Â– as you might expect, itÂ’s
harder to get users to start a tutorial than it is to get them to complete
First step conversion rate: 71.4%
Additional step conversion rate: 95.06%
Overall completion rate: 37.9%
Many companies are now utilizing the tutorial technique, and it clearly
deserves its popularity. Conversion rates of 95% are practically unheard of,
but tutorials appear to be delivering these results.
An Interesting Trend in Visitor Retention
Another thing I noticed was a strong trend in retention behavior. There are
some remarkable similarities in the pattern of visitor retention across
games, despite the differences in the actual numbers.
Before I go any further, hereÂ’s a quick overview of the concept: Visitor
retention is the percentage of visitors who come back and interact with an
application after their first visit.
Visitors are chunked into groupsÂ—also known as Â‘cohortsÂ’Â—and then analyzed
based on the the behavior of the group as a whole. The most common method is
to group by visit date. For example, one group might consist of all the
visitors who were first seen in the week starting May 3rd.
Once you have grouped your visitors, you can track them over the following
weeks and see how many from each cohort return to the site.
Now letÂ’s look at some actual retention numbers for a variety of different
games. To compile this data, I first took a sample of the different social
games using our service. Then I looked at the average week-over-week retention
for each game.
HereÂ’s a graph of the average weekly retention rates for the different games:
You can see that on the surface, the retention numbers are pretty different Â–
some of these games have long-term retention rates close to 50%, while others
rapidly approach 0%.
However, the interesting thing to note is that while the absolute retention
rates are different, the pattern of retention is very similar across games.
They all have a massive dropoff after the first week, with relatively flat
retention in the following weeks. If you take a closer look, the Â‘flatÂ’ parts
of the graph run nearly parallel, meaning they have very similar weekly
We can take a closer look by calculating the Â“conversion rateÂ” Â– (e.g. week 3
divided by week 2, etc) between adjacent weeks. HereÂ’s a graph with this
See a pattern? At the first point on the x-axis (Week 0-1), we can see that
the initial conversion rate ranged from 1.76% on the low end to 62.83% on the
high end. The interesting part comes later, though Â– no matter what the
initial conversion rate between weeks 0 and 1, the following weeks convert at
close to 80% across all of the games.
Basically, this means that once youÂ’ve had a user for at least a week, they
have an 80% chance of coming back each following week.
This suggests that your initial retention rate is critical, because once
youÂ’ve retained users for a week you are likely to keep them for quite a
while. This behavior also raises another question: why do almost all of the
games in our sample exhibit this behavior? Is it possible that this is just
how social games work Â– retained users have an 80 Â– 95% chance of returning
each week? If so, this could mean that the only thing you have control over is
the initial retention rate. Time to write and polish your tutorials.