InsureMe Affiliate Blog: A/B Testing Made Easy

The PPC team here at InsureMe spends a significant amount of time on A/B testing. And we're often amazed at the results. Often, the ads we predict will have the highest click-through rates don't—and those we don't expect to perform, do.

That's why A/B testing is so important to start with. By changing variables and closely monitoring the results, we find what works for each particular ad group—and it's not always the same from one ad group to another. But the more we learn, the more successful our PPC efforts...and that makes it all worthwhile.

It never ceases to amaze me what draws consumers' attention and compels them to click on an ad. Sometimes it's all about the wording, other times it's the total message. Regardless, making our ads stand out by being different from our competitors' seems to please both the search engines and the searcher.

In performing ad testing, I've learned a few things I thought might help those of you who are new to the concept, so I'd like to share them with you now. If you've experimented on your own and found ways to help your ads perform better, we'd love it if you'd share, too!

Here's what seems to work for us:

Figure out what makes you different from your competitors, and use it to your advantage. Highlight your business's unique qualities, testing different language, word sequences, text line order and anything else you can think of.
Test one thing at a time. If you try to do more, your conclusions may not be accurate because you won't know which factor caused your results to change. It's best to limit yourself to two or three ads for easier, less complicated tracking.
Make it routine. Ad testing should never be considered a one-time effort. The more you do it, the better you'll get at it! We have a set time we experiment and examine results every week, but you may want to do it more or less often. Once you gain a more thorough understanding of what works best, you'll save yourself the time and money you spent learning in the first place.
Isolate high-performing keywords. Since these are the terms that make you big money, it makes sense to concentrate your efforts on refining these ad groups to make them even more profitable. Doing so also helps you determine whether or not your ads' performance is truly based on these keywords and isn't being influenced by other terms in the same ad groups.
Follow up—and follow up again! Though strong performance history definitely helps, as other factors change so may your results. So check and recheck the ads you test, and make changes as appropriate.

There you have it, the novice's guide to A/B testing. :) Good luck in your own testing efforts!

Comments

You're so right, Penny. Unfortunately this is probably the worst bad conscience of any web advertiser.

I'd like to add two things:

With today's search engines, the price you pay for a particular ad position is directly related to your click-through ratio. Simply speaking, the search engine optimizes its own profits by favoring advertisements that are more likely to generate revenue.

As an example, an ad with a max bid of $0.50 and a CTR of 4% is likely to be listed before an ad with twice the max bid but less than half the CTR.

Increasing your CTR is a much more profitable way of increasing your ad position - and traffic - than just increasing your bid.
The REALLY difficult A/B testing challenge is the landing page. Having several versions of the landing page is so difficult for most affiliates, but even small things like an increased relevance related to the search term and a better call-to-action can double the profitability of the landing page.

I'm beginning to think that the only way to solve this tough problem is to design landing pages specifically for A/B testing already when you create them. It's really the only way to make sure you do this instead of something else.

Posted by: Magnus Wester | March 6, 2007 08:42 AM

Hey Magnus, thanks for the input! You make some great points about the importance of high click through rates and great landing pages.

Since most affiliates don't have the resources to pour out endless streams of money into their PPC campaigns, some hard work and experimenting on ads to raise CTR can really pay off.

I'm a huge believer in testing landing pages, too. Though it's difficult for affiliates to find the time to create landing pages specifically for testing, I do think you're on to a great idea. I'd recommend affiliates spend a little extra cash--even if only once in a while--on the services of a professional writer, to have great quality copy they can use for testing. Someone experienced in search engine writing can contribute a huge amount of expertise at relatively low cost--especially when you consider the difference it can make in conversion and sales over the long term. Once you've used some of these well-written pages to find out what kind of copy really sells, you can use what you learn on many future campaigns...and that just makes good financial sense!

Posted by: Penny Hagerman | March 6, 2007 10:34 AM