Split-Testing Tips for Banner Ads

Posted by Admin on April 17, 2013
0 Flares 0 Flares ×

A/B Split Testing

A lot of advertisers write off a campaign as a failure after creating a handful of ads that don’t deliver the CTR or conversions they want. Before you give up on those campaigns, have you tried A/B split testing your ads? This means testing different variables on your ads to determine which performs better.

A successful split test consists of changing one variable at a time. There are many elements that can be tested but here are my top 4 recommendations:

The Header – changing even one word in your headline can have a significant impact on your overall conversion rate. For example, does asking a question yield better results or stating a fact/feature about your product?

The Copy – play around with the length of your wording, font, size, bolding and colour. You can also try inserting the word “free” or offer a promotion or some other time sensitive incentive to increase interest.

The Image – test professional vs. amateur images and images with people vs. product alone.  Does a border around your image make it stand out more? What about black & white vs. a colour image?

The Call-To-Action – try actionable words like buy now, order now, sign-up/subscribe, download and share on social media.  Test out buttons that are different sizes and colours.

AdChop published this cool Infographic on Ideas For Split Testing Your Ads – some suggestions are far out hilarious but don’t laugh until you’ve tried them out!


The Strategy:

Test three to four different variations per group. Anymore than that and you risk diluting the traffic and unnecessary spending.

One school of thought is to test radically different banners – think macro to micro. As a winning ad emerges, you can then tweak and make minor variances.

Keep ad conditions similar such as bidding as close as you can to the same amount, testing at the same time of the day and for the same length of time.

Focus on one traffic source at a time and set a budget for each round of testing.

Rotate your ads evenly for accurate split testing.

Make sure your landing page is consistent with the messaging on your ad. There is nothing more annoying that clicking on an interesting ad only to be directed to an irrelevant page.

Before you decide on which ad works the best, typically a minimum test will be 2-3 times the payout of the offer. For example, a dating campaign that pays $10/lead, you will spend $20-30 on each banner. If you are not being paid on a CPA (but you are getting paid hourly, on a fixed budget or on CPM) , you would want to get the equivalent of 100 clicks based on a relevant CTR for that niche and traffic source. For example, an average CTR for a dating campaign is approx 0.1% so in this case, you would test around 10,000 impressions in order to gather enough data.

Don’t disregard a lower CTR too quickly but evaluate the entire output. High CTR also doesn’t always equal high conversion rates. For example, evaluate CTR along with the conversion rate to find out eCPM (earnings for every thousand impressions) or EPC (earnings per click).

Remember CTR is just part of the formula, savvy advertisers take into account both the CTR and conversion rate.

And of course, you can always use our Display Tool to quickly identify the top ads by advertiser or keyword so you can see which ads are performing the best. Get inspiration to use on your own creative and copy so you can intelligently test new ads.

Split testing can be a timely process but with creativity and patience you can get those higher conversions which will benefit you over the long run. These are my tips. I’ll love to hear what’s working and not working for you!


Topics: Optimization, Banner Ads


Leave a Reply