To introduce us to the space of contextual advertising is Joshua Stylman, from Reprise Media. He begins by talking about the value proposition of search. The goal is to connect consumers when they’re interested and engaged. He gave the example of a search for “hotels in Chicago” that within 2 clicks he was at an Expedia page for information on hotels in Chicago. He illustrates the reality of search marketing. Google and Yahoo only over about 5-8% of the targeting, where there are many many other sites that have a more targeted audience which enables contextual advertising to be a growing and huge market.
He goes into the directory driven advertising, these are such places like City Search. There is no technology in there to determine what is the most relevant. Its about navigating categorical content and drilling down to find what your looking for. Josh said that many of these companies realized the contextual ads were not scalable. The problem selling the ads and dealing with the people, it wasn’t working so well. About 2 years ago Overture and Google started to look at how to solve this problem. He mentions several other 2nd tier engines that have jumped into the landscape. He then gives an example of the non contextual ads, such as the weather channel. He did a query for weather in New York. Most of the results are for travel in New York for people that are outside traveling to New York. Not useful for someone in New York looking for weather information. He doesn’t think all the engines are there in terms of providing a broader effort to match the person with the content.
Big channels in 2005, in-content advertising, such as those found on Vibrant Media. One of the major SEO forums uses this same type of advertising. It is basically advertising or links inside the content. If you roll over the ads in the content you get a box with the ad information. The next big channel is in-feed RSS advertising. People can pull these feeds in from a variety of sources, such as Search Engine Roundtable. Bloglines for examples is one program you can use to read and gather RSS feeds together. He mentions that spam can be a problem in this channel. He ends with positive notes on the whole space.
Up next is Brad Byrd, from NewGate Internet, to talk in more depth about contextual advertising. Last time I heard Brad speak he was very good, I imagine it will be the same this time. He starts with some case studies in contextual advertising.
First case study, is looking at traffic numbers from major pharmaceutical company. He looks at % of content traffic, % of all paid traffic, conversion rate, and cost per conversion. He compares the following engines: Google Adsense, Overture Content, Quigo AdSonar, and Vibrant Intelltxt. What is shown is that Google return about 75% of content traffic. The conversion rate is better on Overture for paid search, where Quigo is higher as well with 6.1% conversion. The cost per conversion varies in this case study, it seems for the client Overture is return the best cost per acquisition.
He moves on to his next case study for a web technology company. What is found is that content traffic costs more, but its can be a better vehicle for conversions. They see this and want to reduce the costs the client is paying in the content network. So you discover that content totals are a problem, what do you do? You need to find the problem keywords, but you can’t get rid of just the high cost keywords. He says the solution is web analytics. They find in the analytics that keyword 1 and keyword 2 are generating a good amount of traffic. These will need to be tested.
Key points to consider in Google. You can receive Google URL reports attribute all content traffic to the default URL for an AdGroup, NOT to the URL assigned to the given keyword. Google SENDS traffic to the URL assigned to the keyword. This makes Google reports useless for tracking content performance, and requires tracking tools specifically to track this.
Brad says they find this by studying the target keywords. So what they do is created a parallel campaign, pull out the keywords. You create the new group, add keywords, and make sure the pricing is the same. Content wise they cut the pricing in half.
Some key points from this. Adsense matches content against AdGroup titles, descriptions, and keywords. Smaller AdGroups are better for targeting hot ideas. Adsense fundamentally matches and targets against individual keywords (eg. Dynamic keyword insertion). If building single-keyword AdGroups for content is impractical consider strategies to identify top “content keywords” and break those out into their own AdGroups.
Andrew Goodman was up to present an overview how effective the ads are going and where the space may be going. Andrew says we like search engine marketing because can you really afford to annoy 20 million people? He gives the example of a remote shower spy camera, that no one needs but you could probably sell it online with popups. It’s a terrible type of advertising. The clear results of say Google results are way more effective and clear. He gives an example of a search “portulace seeds”. Contextual ads are not really search, so why is Google and Overture going after this inventory. He says that for awhile there was not enough ads to buy. In the beginning CTR for contextual ads is much lower, but that’s no problem. Conversion rates were the same said the search engines, but what if they are not for you. Someone else conversion rates is NOT for you.
Take home from this: Bid lower for content networks. Leave bids on, but turn off “content targeting”. You may have to create another AdGroup but its worth it. What they found was that the content targeting workaround, worked great. It cost less but conversions stay the same. Significant savings by doing low bids for content targeting. Try it.
Andrew also mentions that they tested image ads contextually in Google Adsense, and they performed terribly. The end he says, they turn them off.. Interesting, there has been good talk about whether these would work or not. He also found that content ads began performing poorly for core words. There is some continued to work well for “tail” (lower volume) words.
He presents some painful truths to end, they are: Contextual ads don’t perform well if you sit on your heels. You must bid carefully! You must track them separately.
Two of the search engines Google and Overture had 3 minutes to briefly talk. Emily White from Google said they had one slide. She puts up a huge slide with Google on it to engrain the brand even further in our heads. ; ) She says that she is seeing an increase of about 50% with conversions since Adsense inception. Advertiser ROI has been steadily improving throughout the year as we have launched targeting improvements and, an automatic CPC discounting mechanism.
Barry Chu from Overture was up next. He works as senior manager of Content Match. He says that Overture is investing heavily in content match in the next few years. Content match is apparently a completely different product than search. Its not the same and Overture quickly noticed this. They want to make sure distribution is appropriate and you can get the leads that you want.