Author: by Christian Fager

[Marketer and indie developer Christian Fager outlines exactly how to use split testing to hone in on your audience via Facebook ads -- driving traffic to your fan page or home page and determining specifically what demographic is most interested in your title.]

If you have never run Facebook ads for your video game, you may be missing out on some very good and inexpensive data about your game's market.

We were finding it difficult to get real data from potential customers. We needed to find who would be interested in our upcoming game, and what parts of the game would draw them in. We wanted a realistic market response, so instead of surveys we chose ads.

There are five key things a Facebook campaign gives you:

My Example Campaign

In this example test, I ran a campaign for an upcoming game Breed Battles. The end result was 110 new fans from 173 clicks, market demographics, which groups they are part of, and what information they were most interested in. The cost of this campaign was $70.30. The ads were shown to a total of 136,556 people an average of 3.4 times each.

Many campaigns generate more fans for cheaper. It highly depends on the topic and how much you want to test. Bringing in video game fans compared to other topics was more expensive than any other Facebook campaign I've run.

Profile Your Customers

If you have no idea who your target market would be, or you just want to confirm who it is, it can be discovered through testing.

Initially we targeted:

The above groups have about 4,839,340 people in them. Those topics were chosen because we thought fans of those topics would like this game too.

The Results

Here is an image of most successful ad of the campaign:

We'll analyze each part of the ad shortly.

Setting Up Your Ad Campaign

When you set up a campaign, you can choose where you want to send people that click. You can send them to your website or to your Facebook Fan Page. If you send them to your website you can control the title of your ad, which is great for increasing clickthrough rates. If you are sending them to a very specific page on your site (such as newsletter squeeze page, sales page, or some way to build your relationship with them more than that one time) then this can be a good way to do it.

If you are just going to send them to a regular home page (that isn't tailored to retain them for future relationship building) then I don't recommend it. You're far better off sending them to your Facebook Fan Page. That is what we chose in this campaign for two reasons:

Those reasons make up for not being able to control the title of the ad unless your website page is very strong.

Creating a Fan Page

If you don't have a Facebook Fan Page yet, creating one is very simple. Simply go to another Fan Page like this one and click the "Create a Page" link on the top right.

That will bring you to a screen that allows you choose what type of Fan Page you want. You can use your business or the game itself (Many games use the "Brand or Product" category and "Games/Toys" from the drop down).

You can then insert your game's profile picture and type something in the "about" section. Now you are at the new Fan Page layout and can set up everything from background, videos, write on your wall, put your website link in, etc. After you have the page setup, invite some friends to get a few likes.

There could be entire articles dedicated to how to set up your page, what images should be on it, etc. The goal is to convert as many visitors to fans as possible.

You want the best user experience possible and to at least fulfill their expectations. If they click an ad that says "If you like bananas you'll love this recipe", and it turns out to be a page about meatloaf, they will most likely click away within five seconds.

That is exactly what happened with one of our test ads.

The ad received a great clickthrough rate and would have been highly efficient, except when users got to our page, it didn't meet their expectations of being similar to StarCraft. Lots of easy clicks, very few conversions to likes.

Make Your First Ad

Choosing an Image

We wanted to split test the image of our ad first. This started with trying six images of various creatures from our game. Our best-performing image, the Sphinx, had a 50 percent higher clickthrough percentage than the number two image in identical ads.

There are a couple reasons for this:

So make sure your image draws attention, stands out, and it's likely to inspire curiosity.

Selecting Demographics and Interests

Continuing with the ad setup, choose what location you want your ads to be seen. We used "United States".

For age, you can use what you think your target market is, or try all ages.

Sex depends on your game. We knew our market target market would be mostly guys, so we had only "men" selected.

When you find out what your target market is, make sure to refine these to best fit that. The more targeted your market is, the fewer wasted ads will be shown (which helps both CPC and CPM, which we'll get into shortly).

Interested In and Relationship Status we left to "All".

Interests -> Precise Interests – Enter in topics you think your market is currently fans of. We started with nearly 5 million people, but later on narrowed it down to ads with 100-300k people.

Education we left to "All".

Workplaces we left blank.

Connections -- "Only people who are not already fans of Breed Battles" (or your page, of course).

Pricing and Scheduling

You have a very important choice here. CPM (Cost Per 1,000 Impressions) or CPC (Cost Per Click). To keep this as simple as possible, it's much safer and easier to use CPC. In the campaign we ran, we tested many variations on CPM, and it was extremely easy to drain cash with less than stellar results.

The ads are not equal. A CPC ad will get a better placement (higher up on the screen) than a CPM ad will. The CPM ads will show up in lower and easily ignorable spots, from a Facebook user's perspective.

For example, we ran the exact same ad on CPC and CPM, and the CPC versions normally received 500 to 1,000 percent higher clickthrough rates. CPM can work if you have something that is very high converting, but it's a lot easier to waste money with (note: this is different than how Facebook previously did things; we used to run campaigns on CPM that were the equivalent of 2 cents per click on a well-designed ad, but Facebook has since changed to this newer system).

How Much To Bid

This depends mostly on how patient you are. If you're not in a hurry for your ads to show up, you can bid less than Facebook's "Suggested Bid". We'll use a CPC range of $0.17 - $0.41 as our example. If you want to see fast results then you may want to bid about $0.25. If you want to get the best price possible you could put in $0.15 and hope they show your ads that day.

The bidding system Facebook uses will play the highest bidding and converting ads. That means if your bid is $0.25 and the other person's bid is $0.21, and both ads convert equally, your ad will get the spot and you will pay about $0.22 cents per ad (instead of your bid of $0.25).

It's highly dependent on how well your ads convert. If your ads take 1,000 people to get a click and a competing bidder's ads only take 300 people to get a click, their ads will have priority over yours.

Start low and if they don't get shown fast enough, slowly increase your bids. If you choose to use CPM, I recommend you start extremely low (if the suggested bid is $0.15 and you bid $0.16, it may take 10 minutes for your daily budget to be used up, and you may not have a single click).

Finishing Your Ad

I recommend you keep your daily budget very small to start. $2.00 can get you an idea how your ads are performing, and you can always raise it later. If it uses up the $2.00 and you want to keep testing during that day, you can raise it, and lower again the next day. When you are comfortable with the ad's performance and want to build fans fast, then you can raise this up to $10, $20 or $100 per day if you want.

Once your ad is ready, you will submit it. Facebook will review it to ensure it meets its guidelines (Common sense applies: don't use copyrighted material. You can read all the policies on the site). Most of the time Facebook approves ads in 12 hours or less. If you make changes to the ad (other than changing your bid) the ad will have to be re-approved, but that generally takes less time than the initial approval. By the end of the Breed Battles campaign, Facebook was approving the changes to my ads in less than an hour. I've read that spending more will help your ads get approved faster too.

The Most Important Thing to Take from This: Split Testing

If you remember one thing from this article, make sure it's this: Split test everything! The picture, initial body of text, call to action. Split test one aspect at a time.

A normal A/B split test runs two ads that compete head-to-head. The ads are identical except one variable. In this example, the only variable between ads A and B is the image.

If ad A converts at 1 percent and ad B converts at 1.2 percent, you keep ad B and make a new ad C (the only difference will be a new image).

The ads B and C will now face off in round 2. You will run the ads several rounds, replacing the loser's image each time. When you find the best converting image you can move on to split testing the next part of the ad.

Let's say you ended up with ad F which converts at 1.3 percent, and you want to move on to split testing the body text.

You now run ad F against ad G (now they both have the same best-performing image from the first split test, but different body text). You will run the winners of the body text the same as you ran the image competition. Now you end up with an ad Q that converts at 1.5 percent.

You take Q, duplicate the image and body text, but will have it compete against an ad with that has a different 'call to action'.

The call to action is important because it tells the user what they should do next. Even if someone is interested in your ad, they are not likely to click it unless you give them a reason to. If you want them to "Click to Try the Beta", "Like This Fan Page", "Visit the Website", "Sign Up For the Newsletter", make sure and tell them! Your clickthrough percentage will be much higher with a call to action. Now, once you're done having your calls to action battle it out, you are left with ad W, which converts at 1.9 percent.

Now you can continue split testing with the landing page, go back and split test the image again, or change up who you are targeting. Either way, those small improvements in clickthrough rates from the original A at 1 percent to W at 1.9 percent are a 90 percent improvement over the original ad. That's almost double the clicks for the same cost!

You can also split test different groups that will see your ad. For Breed Battles we found out Yu-Gi-Oh! fans had a higher clickthrough rate than other groups and topics.

Facebook is a different beast than most split test campaigns, due to the longevity of ads. The longer campaigns run on Facebook, the worse they tend to do (diminishing returns). They have a small shelf life compared with Google AdWords or other service, so you may not be able to split test everything to the maximum before that ad becomes ineffective. To counter the diminishing returns you can...

Multiple Split Test

A great way to find the highest converting ad as quickly as possible is to split test multiple ads at the same time. This is also helpful instead of constantly waiting for Facebook to continue approving one ad after the other. I run about six ads at a time (still only split testing one variable at a time).

When you want to make a new ad, just select a current ad and "Create a Similar Ad".

How to Split Test Accurately With CPC

The one issue with CPC is getting accurate numbers from a split test. Since you are paying per click, Facebook wants to use the ad that will get it the most clicks with the fewest impressions. The site will try all the ads for several hundred/thousand impressions, and usually give one of them a much higher number of impressions because it performs the best.

This will partly tell you which ad of yours is working well, but usually several ads don't have enough impressions for an accurate determination of their clickthrough rate. When this happens, I will adjust the bids, lowering the bid on the one Facebook is over-showing or raising the bids on the ones Facebook is under-showing. The goal is to have the ads have equal numbers over the same time period to get accurate results.

The system isn't perfect and requires staying active with your ads throughout the day, but it gives better data than not doing so.

Analyzing Your Data

After you get a good number of fans to your page you can really analyze the data. In the administration section of your page you can go to "Page Insights" to check out fan demographics, article engagement, page reach, and more.

The reports for the ads specifically are in the Ad Manager, and you can look at how your campaign has performed. It gives you four choices for viewing stats on your ad performances. If these reports aren't showing up initially for you, give it a few days or get more fans from the ads first, it needs a minimum amount of data before they show up.

Conclusion

Hope this helps you with an ad campaign you can set up for your game; much of it can be used for non-Facebook ads as well.

Don't follow every step exactly how I laid it out here; your campaign will be different. The biggest thing to take away is the importance of split testing. If you do that, it will make up for most mistakes you might make.