|
Monday, June 26, 2006
Recently Microsoft launched adCenter Labs, making it available to the public.
It's always fun to play with beta tools. But hard-core marketers will want to know if anything is on this site that can help them with their existing campaigns.
Here are a few notes on features I checked out.
1. Keyword Group Detection. This tool was pretty unsatisfying. It behaves similar to other keyword research tools on the market, and gave me no new ideas over and above what I might find using the Google AdWords keyword tool. In some cases no related keywords were "detected". Food for thought is easier to come by, though. If you quickly click on the related keywords, and click again, you can get from "hard money" to "mazda dealerships" to "puma jamaica" to "rod welding injury" as quick as can be. And this leads us to wish we could do more "broad brush" run of site advertising by going for "super broad match" - at least in some industries. Google's algorithms mostly prohibit this. Editors at Yahoo and MSN will put the kibosh on a lot of terms as well. Which makes using the tools less fun than, say, buying ad time on Will and Grace. :(
2. "Search funnels" is fascinating - making public the data on what else people search for after an initial search. For example: 1.2% of users search for "eBay" after searching for "thong". Another 1.1% search for... well, I can't say. But honestly, people. This is not a frivolous example, but rather, tells you something important about search behavior. It's volume driven. Try something relatively popular like "contact lenses" - there isn't enough data because it's not a popular enough search term. But "thong"? Plenty of info on that!
3. Search mutations. Again, we seem to be waiting on data here. Nothing significant is found for "Derek Jeter" or "Delgado". Nor for the top brand names in some retail industries I tried. But plenty of info on Britney Spears mutations. Which probably makes you think of a joke. Even if you're not a professional comedian, that oughta be a slam dunk.
4. Content classification. I have to say, this one lets a lot of the cat out of the bag on how much search engines know about our sites. It gives a "confidence score" related to how closely the technology thinks the site comes to a pre-set universe of categories. OK, so with an obvious content site like cnet.com, several categories are competing with confidence levels in the mid-teens (15%, etc.). Trying a variety of content sites and homepages for service-based companies, I see a similar pattern. The info appears vaguely accurate. I tried typing in the URL of an ex-client, though. Though the product line *promotes* personal health, the functionality of the site is totally skewed towards e-commerce, i.e. selling the product. So the confidence score for "shopping/online stores" was .993! And "personal health" as a content category was the only other category that was on the radar, at a scant .002. Fascinating. For another client, though, the "shopping/online stores" score was only .250, even though it's just as much of an online store as the other. A third retailer I know came in at .555, as he should. He sells thousands of products on his site. Why the disparity, I wonder? Very tough to say. None of these sites has significant content, though all are being encouraged by the likes of me to develop informative landing pages and to post more supplementary background materials. Even .250 is pretty high, though. Seeing the disparity, it makes you wonder how much of Google's index is driven off relatively volatile predictive scores like this. And how unwarranted some of the differences are likely to be.
The long term trend is likely to be that they'll do a better job of pinpointing the "meta-purpose" of sites. Down the road, I don't expect you'll be able to fool a search engine into thinking your e-commerce site is full of "valuable content" unless it really is. This is not so much a question of what search engines should do about what they know... but more about what they know. They know more all the time, and they can use the knowledge if they wish.
So we're out of time for now. Trying these tools, did I come away with information that could immediately help in my campaigns? Not really. Indirectly though, I already feel edified. The smart money will definitely use these kinds of tools to understand their markets, and to understand what search algorithms think about.
Posted by
Andrew Goodman
View Posts by Category |
|
|
