September experiment: Ad refresh rates

I’ve been running an experiment this month, trying out a reduced ad refresh rate. Bus Jumper used to refresh ads every 45 seconds. Flurry shows that the median session length is 1-3 minutes, which gives me 1-4 ad refreshes per session. I’ve often wondered if I was saturating my viewers with too many ads, which made them tune out, which resulted in wasted impressions and reduced CTR. In theory, higher CTR attracts better / higher paying ad campaigns.

So for September, I decided to cut the refresh rate in half, to 1.5 minutes. This meant that for a large number of games, the player would only see one ad. I’ve had this running since around September 3rd. Yesterday I released a major update to Bus Jumper, which is going to skew the stats. So I’m just looking at stats for about 10 days here.

As you can seem, impressions are a lot lower, as expected. This works out to about 130k impressions for a month, which is about half of what I had before, as expected. The clicks work out to about 1400 for the month, which is a little better than half. So my CTR did go up a bit (1.06% for this period, vs. 0.97% for all of August). Did I attract better ads? Hard to say. The earnings-per-click went from 2 cents to 3 cents. You could say that’s a 50% increase, but that comes with a 50% decrease in clicks, so the two don’t cancel each other out. The earnings work out to about $35 for the month, which is lower than August, which was my lowest month yet.

Now, I fully realize that 10 days is too short a time to draw any conclusions. Maybe it takes time for ad campaigns to notice that an app has improved its CTR. Maybe September is just going to be a bad month. Still, this was an experiment, and based on this data, now I’m thinking that 1.5 minutes is too long. I’ll probably change it to 1 minute for the rest of the month and see how that works out.