How Will An Increasingly Semantic Google Sell Advertising?
By Tom Foremski - May 12, 2009
I went to Google's biennial Searchology event this morning. This is where top execs unveil their latest advances in search.
This year there was a lot of new stuff, and a lot of it revolved around understanding the semantics of search, to understand the meaning of what users want, rather than what they typed. If you misspelled something, or used a seemingly cryptic search term, GOOG execs said it was their responsibility to figure out what you meant by that.
For example, a search for "sfll playoffs" could mean a misspelling of "nfl playoffs" or, more likely, "San Francisco Little League playoffs." Google is using location data and other factors as part of what it calls "rocket science" search technology to try to determine what the user really meant -- not what they typed.
Udi Manber, vice president of core search, spoke about how difficult it was to determine the semantics of user queries, how to understand what was meant.
The demos they showed were impressive. But, how will Google sell advertising against an increasingly semantic set of results? Currently, Google sells keywords to advertisers, and those keywords trigger advertising served up on the results page.
Keywords are very specific. How will this work when the results that people want, aren't related directly to a keyword they typed? Is there such a thing as a "semantic keyword?"
If Google tries to guess what the advertiser wants and what the user wants, it creates more opportunities for a mismatch. Since Google gets paid on clicks, that could be a problem if it misses the intentions of both user and advertiser.
How do you sell "semantic" advertising?
I did ask: as it becomes more semantic in trying to understand search queries, will Google sell semantic keywords? Mr Manber seemed puzzled by my question and said that "semantic" meant many things to different people. I noted the irony that "semantic" has many meanings, especially when Google had just presented technology that it said was good at guessing the intended meaning of a user query.
I asked Danny Sullivan, the search expert at SearchEngineLand, who was sitting right behind me, about this. He said I shouldn't expect Udi Manber to know anything about selling keywords, that he was focused on search. He also said that Google already tries to make some obvious guesses around keywords such as "travel."
Still, it's an interesting question, because there is a lot of chatter about the "semantic web" and "semantic search" in the industry. How would you sell "semantic advertising?" Can there be a "semantic keyword?" How much ambiguity can you quantify and sell to advertisers involved in search engine marketing programs?
It's all part of a larger question, how would you monetize a "semantic web?"
- - -
Please see:
Live Blogging Google Searchology 2009
Live From Google: Searchology - washingtonpost.com
Live Tweets from #Searchology.
« Newswatch: Google Discovers Radio is Not Like Web -WSJ | Main | Journalism Schools Wake Up To Need For Media Engineers »
May 12, 2009 | Permalink | Comment | Category: SearchWatch | Subscribe to SVW
- Top Stories:
- Technology In The Service Of Humanity
- The Death Of The Search Algorithm? Techmeme Has Six Editors
- TEDxSF - Little TED Just Like The Big TED
- SNCR Research: Social Media IS Influencing Business Decisions
- What's Next? Beyond Real-Time...
- PearlTrees: A Novel Approach To Human Mapping Of The Internet
- MediaWatch Analysis Part II: Google Has More To Lose Than Murdoch
- MediaWatch Analysis: Murdoch Will Negotiate Payment For Access To Basket Of Content With GOOG et al
- WeekendWatcher: The Sheer Number Of Things Will Devalue Them
- ChipWatch - Where Will The Next Generation Of Engineers Come From?
- Public Healthcare Could Cut Startup Costs And Help Spur Innovation
- Is GOOG's $750m AdMob Buy Strategic Or Dumb? An alternate view...
Comments (6)
If Google can get search users habituated to a new kind of search results that are related to their keywords in ways they might not have anticipated, that's another opportunity for Google to sell those new clicks to advertisers. Keywords remain keywords. It's the search results that Google delivers in response to those keywords that searchers will finally click on, no matter what techniques Google uses to match keyword search with search results.
Posted: May 12, 2009 5:06 PM
Doug: Fair point but how will Google sell those clicks to advertisers? The advertisers have to choose to buy those clicks in the first place. Will they buy "meaning" rather than keywords? Or, will Google map "meaning" to a keyword? That's a bit tricky and risky in that it might get the match wrong but that could also improve click throughs on all those "enigmatic" searches that it had no clue what ads to serve.
Also, as Google gets better at giving searchers the best result, this encourages organic search or SEO - (no money for Google) over search engine marketing ($$$ for Google).
Posted: May 12, 2009 7:33 PM
It's all part of a larger question, how would you monetize a "semantic web?"
Thats the first thing I am going to ask WolframAlpha.
(hang on, how are they going to monetize WolframAlpha? -ahh, it probably asked itself)
Posted: May 13, 2009 5:22 AM
One of the advantages of bidding on keywords is that buyers know exactly how many clicks they get per keyword and can do analysis to figure out how much they make on their own end. Buying concepts (call them "semantic keywords") would put another process on this layer: you'd need to know what keywords were associated with the concepts you bought to make a good analysis. I think that this is interesting, but requires some serious work on the Google reports the PPC model to users of the system.
Posted: May 13, 2009 11:33 AM
In the near past, if you weren't thinking ahead you were being left behind. Now, if you're not thinking ahead of the folks who are thinking ahead you will be left behind. So, thinking ahead of the semantic web, in the not-too-distant future, a matter of 5 years or so, most people will be talking to some kind of web agent which will then gather the most relevant content tailored for that user from multiple sources on the web, it will then gather ratings and gossip and whatnot to help you make a choice. If not for the rise of open source this content will be totally controlled by the big players because they have the deep pockets for developing the apps. So depending on whose app you have, (ATT, Verizon, MS, Apple, Google, etc...), and how you use it, your agent will retrieve varying results with varying sponsors making it difficult for marketers to put all their eggs in one basket such as Google. For those who are using open source agents it will be even harder to out-market your competition. You'll just have to be a great company and great things, get great ratings, get great press, have great fans, and network, network, network. Good old fashioned hard work and great service will be the norm if you are to survive.
Posted: June 16, 2009 1:36 AM
Great points Rick.
With the failure of advertising it will all be about your network, who you trust and word of mouth. If you have a lame product or service, there'll be nowhere to hide.
Posted: June 16, 2009 8:20 AM