Day 2 of SES was kicked off with an informative session on ad testing and the importance of setting the tests up right in the first place.
Anton Konikoff, Founder and CEO of Acronym Media was the first to present what he calls “Keyword Driven Marketing”. The concept is pretty straightforward – as marketers we need to decode the customer’s intent behind the search and match the keyword and subsequently the ad to this intent. Sounds easy, right?
Just measure your click through rate (CTR) and maybe conversion rate to see if you guessed intent correctly. Wrong, he talks about the need to use CTR as a process metric and not an end metric and that conversions are what should really drive your ad text testing. He also brought up a good point about CTR in general, that oftentimes CTR can be low if you are qualifying your traffic further. At Acronym, they set up tests based on budgets, so smaller budgets get fewer ads but they make sure to allow enough time to make the test worthwhile (at least 100 clicks, but ideally 1,000) They often test elements such as: punctuation, capitalization, proper case versus sentence case, accent usage and dynamic keyword insertion. He gave some compelling case studies from Sirius and Four Seasons Hotels, where they tested messages such as “save $30” versus “first month free” – “save $30” won.
Nico Brooks, Director of Search Technology at Atlas was next to present. Nico talked more about the importance of testing branded versus non-branded ads. I’ve talked with many clients over the years that are skeptical of the importance of including branded keywords and branded ads in their campaigns. They believe that if their organic listings are #1 and #2 then they don’t need to pay for placement. They usually come around once we discuss that branded campaigns are there to support the organic results (among other things). Nico reinforced this notion and also said that branded ads can increase interest by holding more real estate on the page. Then he recommended using the non-branded ads to create such interest.
Jonathan Mendez, Founder and Chief Strategy Officer of OTTO Digital and http://www.optimizeandprophesize.com/ blog, was next to review the structure of a test. He said tests should be designed around a single question, have a large enough differentiation between testing elements (tap water versus sparkling water is a good choice, but not tap water versus Coke), account for changes in temporal behavior or behavior that may change over time or that is seasonal, collect enough data to reduce margin of error, segment results to provide added insight (Google versus Yahoo versus MSN) and that you act on the results.
Gord Hotchkiss, President and CEO of Enquiro was next on the panel. Gord said that he tossed his presentation at 8am this morning and changed his presentation to address where he thinks ads will be going in the next few years. He said ads will become richer but will have to continue to match searchers intent and not fall to the wayside like banners did. Gord reviewed his heatmap studies from yesterday’s session, but also presented a mock up of what he thinks a Google SERP might look like in 2010, which looked a lot like Ask’s new format. He said that universal search will be mixed with personalization and the example he gave was someone looking for a camera and being shown local ads, image ads, and video ads, and a map. Pretty cool layout.
Finally, Clay Bavor, Product Manager at Google talked about a few additional testing techniques. He reminded us about making sure that ads are useful and relevant along with a good landing page. He also talked about a new report that Google released a few weeks ago where you can now find which keyword phrases your broad match terms are being matched to – look for the Search Query Report.
That’s a wrap from this session – 2 more to go!