A months before we released Runes of Avalon game, I decided to create a dedicated website for it with a simple domain name: www.runesofavalon.com. The purpose of this move was to be on top of search engine rankings when our game is released. As you may know, it takes a lot of time for a new website to break into top10 in the Google search results. We did succeed with this one and are listed at #1 for Runes of Avalon keyword.
The site doesn’t get much traffic, though it is still important to make the best use of it. Once I installed Google Analytics I was able to track my site conversion to download (SR). I was quite shocked that SR can be so low, especially for some sources of traffic. Average SR for month of September was only 25,31%. I decided to improve it. I succeeded with 31% increase.
Since most of visitors enter main page first it was important to improve this one at the beginning. I created an experiment in Google Website Optimizer. It was one of my first experiments, so I know I could do better with selecting what to test.
I created 4 sections on main page:
- bullet list
At the beginning it was hard to think of how I can improve it, since it does look great, but finally I found 2-3 combinations for each section. I was wondering, who is better copywriter: Big Fish Games or I? Description and bullet list was a test of text that are available on Runes of Avalon BFG page. BFG uses very little text as a game description and only 3 points on bullet lists. I had about 10 of them, as I always thought that the more the better.
After first paragraph of text I had two the best screenshots from my game, but it did happen that both of them had been from core gameplay. So I tried to make a mix of one core and one mini game.
Which combination of screenshots is the best? In my opinion the first one, but you never know unless you test it.
I started the experiment of September 19th and ended it on October 9th. Here are my results.
As you can see most of combinations received less than 50 views. From statistical point of view this experiment was not representative, but I decided to stop it before I receive final results. Some combinations were just poor and I didn’t want to make SR lower than usual.
It was a shock for me that BFG description text was worse than my own one (I can’t even speak English properly enough). The next shock was that short bullet list is better than long one. I was also surprised that cutting off one review improved SR.
Numbers on the left: 0/5 say how trustful the metric is. As you can see, I should not trust results and wait longer (but as I said I didn’t want to).
While each section on it’s own doesn’t make big difference, combined could improve site conversion by over 74%. Google showed that Combination 9 is the best one. It was surprising for me because #9 was a mix of sections that didn’t work well (BFG description, screenshot from puzzle minigame, original reviews, original bulletlist).
Control test proved my suspicions. Combination 9 was worse by over 32% than original one.
I stopped the test after 5 days and made another control test with combination 14th. This one was better by over 91%. Google ended this test automatically and decided that combination 14 is the winning one.
Do you want to know which set of screenshot was the best one? The one with a ring minigame.
For reviews section it appeared that less is sometimes more. Maybe those 3 reviews that I quoted sounded just to laudatory. Or maybe it was easier for visitors to find Download Now button (though I have one on top and one right after reviews section).
The problem is:
- the first control test original combination had site conversion ratio 40.8% +-7.2%
- in the second one it had site only 20.9% +-8.1%
Main experiment shows similar conversion for original combination as the second control test. The first control test was probably affected by traffic spike. I will probably make another control test to check which one is better: 9 or 14 and then decide.
What matters is that in the end my site conversion to download increased by 31%. It’s still not as high as I would like it to be, but it always make my marketing efforts more effective.
If you want to find out how to do tests like this with a little effort and read results of my other experiments then come back later to this blog.