Thursday, March 15, 2007

Optimizing for Quality Score Based Ad Ranking

As you've probably read in constant news reports and blog posts Yahoo released algorithmic based ad serving on its ad network. This has been available at Google for quite some time.

The idea behind algorithmic based ad serving is that many factors are used to determine how your ad should rank. These factors include bid, click through rate, conversion rate, ad copy, and landing page copy.

So why wouldn't search engines just use the highest bid to rank ads? I often use the following example when explaining this to people....

What is better for Google, to rank an ad with a $100 bid which has a 1% click through rate or to rank an ad with a $2 bid with a 100% click through rate? If they rank the $100 bid with 1% CTR they'll earn $100 for every 100 impressions. If they rank the ad with the $2 bid and 100% CTR they'll earn $200 for every 100 impressions.

By ranking the ad with the higher CTR they earn more money and provide more relevancy for their end users. Of course the real algorithm is much more complex than this, but you can see the basic logic behind algorithmic based ad ranking.

When you're looking into optimizing for algorithmic based ad ranking it's important to know the impact of making changes to certain aspects of your campaign. Since I am most familiar with this from a Google perspective I'll explain how things work over there. From my understanding most of these rules will apply to Yahoo; however, I make no guarantees.

Destination URLs

The three basic levels you can set destination URLs are keyword, Adgroup, and Ad.

Action: Changing Keyword & Adgroup Destination URLs.

Effect: The next time the Google ad bot crawls your ad it will go to your new destination URL and use on page factors to determine if your quality score should be changed. If the landing page has not changed then your quality score will not adjust and your ads should continue to rank as they have been.

Optimization: Try to set destination URLs on a keyword or Adgroup level. Only update destination URLs when absolutely necessary. If you need to implement new destination URLs create a test Adgroup.

(Test Adgroup: Pause the Adgroup you want to make changes to. Create an identical Adgroup and change the factors you want to test. If everything tests out okay update the original Adgroup and un-pause it. Pause or delete the test Adgroup.)

Action: Changing Ad Destination URLs

Effect: Your ad will be viewed as a new ad and all previous CTR history will be lost. The algorithm will still use factors like your keyword and Adgroup CTR history to help determine your quality score and your rankings, but you will lose data associated with the ad itself.

Optimization: Avoid changing destination URLs on the ad level at all costs. If you need to update the destination URLs on the ad level create a test Adgroup.

Changing Copy

Updating your campaign may also include updating your ad copy or landing page copy.

Action: Updating landing page copy or changing landing page.

Effect: Any changes made to your landing pages will change your quality score and possibly effect your rankings.

Optimization: Use a test Adgroup to check the effectiveness of your landing page changes.

Action: Updating ad copy.

Effect: All CTR history with the ad will be lost. CTR history on the keyword and Adgroup level will still be used to rank your ad.

Optimization: This can be the most frustrating aspect of optimizing for algorithmic based ad ranking. Even if you make minute changes to an ad like adding an explanation point or comma your ad is effectively starting from scratch. This can be especially problematic if you have to change a certain aspect of your ad from time to time.

Say for example that you are advertising a piece of software. Your ad reads "Comes with 5 features". As you ad features you need to change the ad accordingly ("comes with 6 features, comes with 10 features, etc.). Each time you change your ad the ad starts from scratch and you may have problems regaining ranking.

In these types of scenarios I like to create neutral ads in each Adgroup which don't include the variable ("comes with lots of features!"). When I update the ads that do include the variable I have to start from scratch on those ads; however, because I have a variable neutral version I have an ad with CTR history to help me rank for my target phrases as my variable based ads start to build CTR history and rank on their own.

Understanding how changes to your campaigns effect your ad rankings is an important part of optimizing how you manage your search CPC accounts. The key to updating any campaign is the test Adgroup strategy I described earlier in this post. Pausing a successful Adgroup and creating a test Adgroup allows you to back out of any changes and go back to the more successful model at any time.

No comments: