The Dojo is the official blog of the marketing software company TraceWorks.

A/B Testing - We Do And We Don’t.

Morten E. Wulff Mar 16th

An Online Marketing Manager at a new and very hot phone company asked me yesterday in London something in the line of: 

A/B testing is an important way to optimze performance in Search Marketing … How do you do this with Headlight?

(I’ll actually end-up answering this question further down)

Well, first of all he’s right that A/B testing is important. With A/B testing you determine the right ads for the right keywords. This will improve your CTR% which is key to lover your CPC (while keeping POS) over time while increasing your ROI% bla bla bla. The method and theory is pretty straight forward. But when I come to think of it; all the marketing managers that I know don’t use A/B testing. Why is that?

The reason is obvious: When your are working with hundreds or thousands of keywords and ads A/B testing becomes Godzilla of complexity and time killing. Most Marketing Managers and Search Managers are already facing quite a huge job in simply doing the basics; keyword research, copy-writing, and finally uploading all the stuff to Yahoo/Google/MSN.

In many cases optimization is reduced to just deleting keywords that are not performing in terms of CON%, ROI%.

Houston, we have a problem.

Well, the actual question was: How do you do this with Headlight?. My answer was: “We do and we don’t”.

With Headlight we have been working on a slighty different approach - than the traditional process. We think of course like everybody else that A/B testing really sucks - that is the manual process of doing A/B testing. Not the purpose of A/B testing itself = locating the best marketing assets or the right combination of marketing assets. It’s fundamental and powerfull.

(BTW in Headlight marketing assets mean 1. Keywords 2. Ads (text ads / image ads) 3. Landing pages.)

We don’t introduce or plan to introduce any easy (or hard) way of manually doing A/B testing with Headlight.
We do introduce some very simple ways in order to apply automatic A/B testing rules to your campaigns.

Actually we don’t call it A/B testing but business rules. What the business rules do is basically A/B testing automatically across all your selected marketing assets (ads, landing pages) in order to detect the most succesful combinations. 

The business rules was primarily developed to automatically optimize traditional banner campaigns (served either 1st or 3rd party) on request of Henrik Lykkesteen of Saxo Bank A/S. After being asked about A/B testing regarding keywords and text ads I’ll think we’ll ínvest some time and energy into making this possible with Search Engine Campaigns as well.

We try and do it smart … by doing it for you. Hopefully it turns out just as good as it sounds.

3 Responses to “A/B Testing - We Do And We Don’t.”

  1. Jimmi Bram  March 16th

    I have no doubt in the advantages of business rules; the hard part is to make them work efficiently. But if anybody can, we can 

    Every target group reacts differently to rotation. You actually have to consider rotation pool size, length of test period etc. for each target group.

    It is the optimization of the optimization rules, that will really rock the business rules of Headlight.

  2. Wulff  May 2nd

    I just talked to one of our clients in SE space saying that “negative keywords” should be included as an important variable in our optimization algorithm.

  3. Wulff  May 2nd

    I think we need to talk to Saxo Bank on stock market prediction algorithms as well … I think they can help us figure out some clever things on optimization

Leave a Reply