Quest For A Better Quality Score (Or Is It Click-Thru Rate?)

Posted on by John Barron

So I thought that I knew about Quality Score (QS). I choose targeted keywords in a campaign, match those keywords in the copy in an ad and point the ad to a relevant landing page. Simple. I should have a QS from 7 to 10. No problem.

A client of mine, a massage therapy school in Sarasota, FL, asked if I could create a locally targeted campaign to help capture local searchers who were not specific in their search query about location. I already had a traditional (not locally targeted) campaign set up that utilized terms with the city in the term (“Sarasota”) and I was averaging a QS of “7″, so I did not think that it would be a problem to replicate that.

I started by setting up the campaign with the same terms that I had been using in my other traditional campaign with the city (“Sarasota”) removed. The ads had the words “massage”, “therapy” and “school” at least twice. Once I had about a month’s worth of data, I reviewed the Adgroup to determine how I was doing.

I didn’t even have QS of 4! Most of the terms were 3 or under and the terms that I expected would do the best were 2′s. What went wrong?

Plenty. First of all, many of the terms that I was using were no longer as relevant to the ad copy (ex.:”career in massage”). Secondly, I wasn’t using exact and phrase match on some of the most important terms, so the broad match terms were causing ad impressions for even less relevant terms. This meant that the Adgroup’s overall CTR was very low (a big No-No). Lastly, I wasn’t taking into consideration how the account overall would be affecting the CTR and ultimately the QS.

I needed to do some very intense intervention. The first thing that I did was review the keyword list for terms that were less relevant. The terms must have the words “massage”, “therapy” and “school” in them. Plurals were okay, but if I didn’t have two of those three terms in the targeted keywords, I would pause the keywords. Since I had many terms that had “career” in them, I just decided to make another Adgroup focused on targeted keywords that included “career” in them. The second thing that I did was change some of my broad match terms to phrase match. I knew that the terms should be relevant, but keeping them broad meant that there was a lot of garbage that was causing my ads to show. Thirdly, I used the Keyword Report (under “See Search Terms” at the top of your Keyword interface on Google) to look for negative keywords so that they can be eliminated. Lastly, since I began to see improvement, I decided to use the same approach on other Adgroups because CTR for the entire account influences QS.

The result was an enormous improvement in CTR for the Adgroup (from .29% to 1.59%). It also led to an improved QS (most going up one or two digits) and a lower cost for the client (by about 15% per click). And best of all, better CTR usually leads to more conversions.

Here is a summary for QS improvement:

  1. Make sure your Adgroup is limited to terms that are very relevant to your ad copy. You can always make other Adgroups to better target your terms.
  2. Change broad match terms to phrase or exact match. As a note, pause (not DELETE) your old terms. Deleting them will eliminate valuable data.
  3. Look for negative keywords.
  4. Apply these approaches to other Adgroups so that your account improves overall.

I hope that this helps. Let me know if you have other questions.

Google Behavioral Targeting: Not Quite Ready For Prime Time

Posted on by John Barron

In all of the hullabaloo about Google+, there was a smaller announcement on June 23rd in Search Engine Land about the public launch of Adsense’s Behavioral Targeting tool. Google had been perfecting the tool through a beta launch for over two years. I had been eagerly awaiting the service since it was first introduced as a concept many years ago. The system would profile users into interest categories based on sites the user has visited. There would be three criteria: type of site visited, frequency of visit and recency of visit. According to the article, Google has over 1,000 categories.

1,000? Sigh.

There should 10,000…100,000. This is Google. Big numbers come easy. Netflix, using Cinematch (their proprietary recommendation system) apparently claims “predictions are accurate within half a star 75 percent of the time” (How Stuff Works). If Netflix can develop a system that is so accurate for predicting movie choices, Google should be able to develop a more sophisticated category interest system.

Now, this could be that Google is treading lightly so that it does not run afoul of privacy rules. Google and other ad networks have worked very hard to comply with rapidly evolving regulations. If a crafty researcher could determine the identity of a person based on these interest categories, Google is in trouble. It happened to Netflix.

If you have a gmail account, you can view your interest categories (this is the fun part!). I will share my interest categories with you to further this discussion. Don’t judge me.

google interest categories

I should list the things that are good first. It shows “Bicycles and Accessories” as an interest and I am an avid biker. I was looking at a FB page of a favorite biking resort earlier today, so the list might suggest that it classifies very recent behavior and I am an ideal target for biking-related ads, but it was a travel destination, so not 100% perfect, but still good. It may have been from another source that I was unaware of, so Google knows me better than myself :) . My age (25-34) is wrong (and I am not telling you my age). It could be because my daughter was using my computer, skewing my age, but it is off and could use some help. The real estate listings interest categories are good targeting, but couldn’t they define a region, city or state? All of the movie interest categories are too generic. If they could learn about the movie genres that I was looking at (adventure, sci fi, fantasy, etc.), that would be much more helpful to the marketer.

Now, if I were Netflix, I would target this profile, but how is this different than targeting a online publication like imdb.com?

This information is a great start and worth exploring further, but so far, most of the categories are too broad to be helpful and require refinement. Note the opt-out button. I haven’t seen this prominently advertised throughout Google, but that is another post. What do you think of Google’s interest categories?

The Display Network: When To Use It

Posted on by John Barron

It is unfortunate that the Gods at Google automatically assume that an advertiser wants to place ads in the Display Network. Many advertisers would probably feel that there ad dollars were being wasted if they new what the default setting meant. It is not as if I feel that the Display Network is a bad investment or misleading, but it is extremely sophisticated and many advertisers lose a great deal of money never knowing that their ads are not on search result pages at all. Google ends up looking bad and the advertisers lose faith.

The Display Network can show Google text, image, video and mobile ads across Google’s content network. These ads can be shown in an automatic format where Google selects websites for you for ad placement based on your keywords or you select certain web publishers. When handled well, these ads can pinpoint a target audience very effectively. Typically, however, they are only an afterthought and result in poor traffic.

The Display Network can produce great results, but it is not a search ad and should be treated differently. The audience is in a different mindset and the message has to work harder to attract the user.

Here are some suggestions as to why you should consider the Display Network and how it may actually perform as well or better than Google’s search network:

  • A New Product or Service: Let’s face it, someone has to know about a product or service before they can search for it. An informative Display Network ad could entice a user to want to learn more.
  • Different Audiences Using The Same Search Queries: There are times when you have to reach out to an audience specifically and you can’t run two different text ads that will appear at the same time (ex. long term and short term RVers). You can run two different ads, but the different ads would ignore one of the groups. The answer could be to have the search text ad focus on the more profitable group, while the display focuses on the less profitable group.
  • Reaching An Audience Before They Start Searching: Ian McAnerin, of McAnerin International, once told me a story of trying to attract event planners to Canada. He originally targeted them on the search network, but it seemed that when the planners were searching on Google, they had already made up their minds about where to go, they just needed details. He then switched to the Display Network so that he could reach them before they made a decision. He probably picked trade sites, destination sites and business pubs.

The Display Network has certainly come a long way and I anticipate that it will even get better when it comes to targeting certain audiences. Approach it with caution, but do not ignore it.

Loving Life at SMS India in Bangalore

Posted on by John Barron
Akshaya Patra School Lunch Program

Enjoying Recess After Lunch!

I am having the trip of my life. Late last month, my good friend, Suresh Babu, emailed me to invite me to speak at a search marketing conference…in India! I have never spoken at a major search conference, although it has been a major goal of mine. I have spoken at business meetings for associations, chambers and other venues, but never a presentation in front of peers.

After the initial sense of denial, it became clear that I would be going to Bangalore. This was an incredible opportunity both professionally and personally. I was thrilled both with the chance to speak as well as to travel to India.

What amazed the most, however, was unknown to me until I arrived. The beneficiary and co-producer of the conference was a non-profit based in Bangalore called Akshaya Patra, a free lunch program serving  children throughout India. It started in 2000 feeding only 1,500 children in five schools. Now it serves over 1.2 million! My first day in India, I was to meet some of the staff and trustees of the organization as well as take a tour of their facilities. The kitchen cooks 40,000 pounds of rice a day, Monday thru Friday. The cooking starts very early in the morning (2 am). The tour ended with a visit to one of the schools where the lunch is served. I was able to take part in the program, feeding the children who were excited to get the meal. The meal not only sustains the children, but gets them to stay in school. The power of the charity was emotional and inspirational. I was blessed to take part in it.

I will write more about the content Search Marketing Summit India in subsequent posts.

Conversion Optimizer for Adwords Makes Sense

Posted on by John Barron

I recently tried the Conversion Optimizer for Adwords on a client of mine. The results were impressive and I would highly recommend it. They were not evenly divided, however. I have two campaigns to which I applied the Optimizer. One campaign particularly targets branded terms. The second campaign is more for generic terms, but certainly converting.

In the case of the branded campaign, there was a 29% increase in revenue when I compared three weeks worth of traffic. This was a great increase to report to the client! The generic terms, however, showed not increase whatsoever.

Because of the improvement, I certainly pleased with the result, but if the Conversion Optimizer knows when and when not to bid based on keyword, geographic location and other performance factors, why wouldn’t I see a greater improvement on generic terms versus the branded terms. The branded terms should have a very narrow focus when it comes to how much potential variance there is where as the generic terms would have a wider range of possible factors (keyword variations, negative terms, etc.). If Google knows better, it should know with greater certainty which terms perform.

It may be that I have not allowed the Conversion Optimizer to run as long as it needs to, so I will revisit this comparison down the line to see if it has ultimately done better with the generic terms. Other blogs that I have read (such as “Google Conversion Optimizer: The Best-Kept Secret in PPC?“) If it turns out that the Conversion Optimizer works well for the branded campaign, I may try the Enhanced CPC campaign on the generic terms.