As I strive to build my own HTML5 game portal I am becoming increasingly conscious of the importance of direct traffic.
Over recent years as my games have found their way to different portals and the games on my own web site have been called up from countless sites, I’ve built up a pretty useful link network. I figured it was about time I properly structured each of these pages with analytics code so that I could measure performance and try to figure out what was working and what wasn’t.
I use Google’s Analytics.
Since a number of these pages have AdSense advertising contained within them I also linked the Analytics property to AdSense. Some further information on how to link AdSense and Analytics.
I quickly identified that much of my traffic was referred from an external source.
What does this mean?
Simply put it means that the analytics code has recorded that when the page was loaded (and the script executed) it was as a result of a link or referral from a previous page.
A link is obviously a physical click on an hyperlink and a referral may well be as a result of a redirect which can be achieved in a number of ways. Either way the visitor arrived at my site from an external source.
This is good and healthy and all that but I’d rather people came direct to my site as a result of them liking what they’ve seen and using a bookmark.
So I started to scratch my head and consider how I might encourage direct traffic.
I put my gamer’s hat on and played a game on the portal. I went through the game experience all the way to the end (game over) and analysed the experience. It felt pretty flat. The game was enjoyable enough but my only option at the end was to “have another go” or tap a back button to return to the external source.
So I created a banner to advertise my portal and instead of simply presenting a “Game Over” label I also presented the graphic. A clearly marked “X” in the top right corner indicated that you could skip the advert whilst clicking anywhere else on it took you to my portal’s homepage.
My analytics code was telling me that most links were coming in to Galactians 2. So I placed the banner at the end of that game. I re-uploaded the game and allowed 24 hours to collect some data.
The results were surprising.
Within 24 hours my portal’s homepage had gone from single figure hits to triple figures.
Specifically 904 unique hits overnight.
Something was working so I replicated the banner advertising my portal in the second most popular game, Spy Chase.
The unique hits started to rise again. I added the banner to a third game, Danger Ranger.
Another quick look at the user behaviour and I could see that whilst the number of visitors to the games was increasing the number of first time visits was falling. On 8th November 81.94% of visitors were coming for the first time.
Finally the number of referrals on the 8th November stood at 6,353. The total number of visitors for the day was 7,261. So referrals on this day accounted for around 87.5% of all traffic.
My next exercise was to redirect internally the traffic that was scattered around my web pages. This fragmentation of games had always driven me mad so I took the opportunity to redirect everything to my portal. I made sure that if the user was going to Galactians 2 they would end up at the portal’s Galactians 2 page. I didn’t want to break the experience up by presenting another click.
By around the 15th of November I was happy that all games were now pointing to the portal. This was excellent as it meant I was finally maintaining a single codebase.
So one month later what kind of numbers am I looking at?
Well the number of unique hits to my home page on the 11th December was 4,400. A rise of 3,496.
The percentage of first time visitors to the site has dropped to 61.91%. Presumably the 20% shift accounts for repeat visitors.
On the 8th December the percentage of referred traffic stood at 44.63%.
My assumption that the decrease in referred traffic amounted to an increase in direct traffic needed to be tested. It seemed too good to be true. Had I skewed the figures by placing so many redirects on the site?
Whereas previously a visitor was coming from Domain A to Domain B where Domain B was the only place I had analytics code installed, they were now potentially coming from Domain A to Domain B (page 1) to Domain B (page 2). Both pages in Domain B had analytics script embedded.
Would the final resting place (page 2) within Domain B register the event as direct / (none) traffic for Domain B?
Or would it, as I hoped, register it as a referral by virtue of the fact that the original source was Domain A. i.e. external.
I found Google’s help pages to be a pretty useful starting point.
On that page you can download a custom report created by Google to throw at one of your views within Analytics. I downloaded the report and ran it and it came back with no issues.
Digging a little further I installed the Chrome extension Tag Assistant. With this installed you can quickly analyse each page within your portal to ensure that you don’t have duplicate tracking code installed. I’m not sure of the impact here but it seems like good practice not to do it.
I’m happy that the figures that I’m seeing for traffic and referrals are accurate. This is rather thrilling as it means that the educated changes that I am making seem to be working.
Update: after a little digging around the web I found some interesting reading on behaviour related to mobile phones and searches via Google since iOS6. I’m not sure how much this relates to my situation. Probably not a great deal. But it’s worth a read none the less.
And the difference in AdSense revenue generated over the first month as a result of these changes?
A rise of 230% on the previous month.