Posts tagged as:

Search Technology

Beginning on February 5, Yahoo will be changing the way they rank pay-per-click ads. The rankings currently us bid price as the ranking method, meaning the higher the bid price, the higher the ranking of a pay-per-click ad. This coming change will include bid price and a quality score to determine relative rank, similar to what is already in place with Google AdWords.

What is a Quality Score?
A quality score applied to pay-per-click ads means that bid price is not the sole ranking method. This means that an ad with a lower bid price could rank above a higher bidding advertiser. While the exact factors involved in the quality score have not been divulged by the engines themselves, one influence is historical performance and performance relative to other ads displayed at the same time. Simply, if more people click on an ad, it can be assumed by democratic process that the ad is more helpful to searchers and could get preferred rating.

The quality score was devised to try to keep the playing field even. In a pay-per-click process based solely on bid price, nothing can stop a large advertiser with big pockets from taking top spots. In theory, with a quality score feature that effect is minimized as a small advertiser could compete with Wal-Mart by strategically picking keywords and writing well-focused copy for customers. In other words, better quality ads help with better placement.

This process has been used by Google AdWords for some time now, and it does make pay-per-click campaigns a bit more of an art form than a simple function of calculating a bid price. If your SEM vendor can’t tell you why your ad is not increasing in position despite bid increases, the quality score is the reason. (Interestingly enough, the quality score feature can also reveal which PPC vendors are truly savvy marketers versus those that practice "simple" SEM.)

Between now and February 5th, it would be a good idea to scrutinize your PPC ad copy. Also be prepared for some wild position fluctuations that week as well.

{ 1 comment }

While I was away on holiday vacation, I read two articles about so-called "Google killers" – an overused catch-phrase that should affect more people as partially ridiculous.

(Consider earlier recipients of the title, Teoma (here and here), Quaero (here), Baidu (here), and even Microsoft’s Live.com (here), and… you get the point.)

And again this month, still none of these new search engine ideas are novel, or novel enough to be called a "Google killer". To me this idea sounds like someone inventing a new elixir and  calling it the "Coke killer”, or an operating system and calling it the "Windows killer." After all, even Pepsi hasn’t put Coke out of business, as thorough as their competition and advertising may be.

This week’s entrant to the long line of "killers" is Search Wikia, created by Wikipedia founder Jimmy Wales. It is intended to be a
"people-powered search engine" that taps into the essence of Wikipedia, meaning user controlled content.

A volunteer force like this has been utilized before and became an albatross for . Besides, a search engine with user controlled content sounds like it could quickly become a spam magnet.

Next to throw down the gauntlet is Powerset. Their innovative technology claim is that a user can type a search query in "natural language." Sorry, Powerset, but Ask beat you there by about six years.

The other problem Powerset faces is that people do not search in natural language sentences anymore, which is part of the reason Ask re-branded away from their "natural language" search feature. It’s still there; they just don’t tout it anymore. Why? Because users have long been trained to search with
keywords. Powerset would first have to un/re-train this now natural behavior, and then demonstrate it’s more relevant than Google. But that’s not stopping one of its founders, Steve Newcomb, from saying that Powerset could "become the next Google."

Lest I be a stick in the mud, I’m not decrying the ambitious attempts of these upstart engines; I’m questioning the rave surrounding them. Such claims tell me that the product is focused on the competition. The best products, the ones that truly succeed, focus on the customer.

Don’t tell me who you want your new shiny search engine to topple in a few years, tell me why it’s useful to me (and my clients) right now.

{ 0 comments }

With the accidental release of the search queries of 658,000 anonymous AOL users onto the web, people have begun to ask questions about their own privacy. While the search records were considered anonymous because they were identified only by number, several sources, including the NYTimes, were able to track down some of the actual people behind the search data.

A person does not have to be very paranoid to wonder if searches could come back to haunt them. The US government has requested search data in the past. How long before search histories are subpoenaed for a divorce proceeding?

It’s actually the harmless searches that are more revealing. If I search on fear of flying therapy, it should be my choice on who I share that with, not AOLs (or Google’s, or Yahoo’s, etc.) There shouldn’t be a database of my fears out there somewhere.

That’s where TrackMeNot comes in. It’s a
Firefox extension that pulls random word combinations from a list of words and submits them as searches, essentially, diluting whatever cookies and log files may be stored on your searches.

While I think this technology is interesting (and actually kind of funny), there is a downside. I’m concerned how a search randomization extension could ruin keyword research.

Keyword research is an extremely effective tool to uncovering what people are actually typing into engines when they are searching for products in your industry. If I have to start wondering if keyword combinations and traffic estimations are truly accurate, then the advice I give my clients could be compromised.

I think it’s an interesting idea to solve the problem — dilute the log files so much that they are worthless. But couldn’t this problem be solved in a much more effective way? Oh, like, the engines realizing there is a privacy concern and not keeping log files for longer than a few days?

I would hope they draw that conclusion before someone starts a class action lawsuit to
make it happen.

{ 0 comments }

…but why you typed it.

A snip from Scoble’s rant:

When I search on “Office Furniture” why is the first thing I see stores? I don’t wanna see freaking corporate info. I wanna know what HUMANS like to use in their offices.

What Scoble just learned is that the difficulty from the search engine side is not matching keywords, but in trying to figure out the intention of the searcher. "Office furniture," while a logical term, is also quite general, thus making it harder to assume what the intent of the search might be. Each engine has an element of mind reading to do.

For example, a search on "red wine" could give a range of results. The engine has to try to determine if you are looking to buy wine, want to learn about wine making, or matching types of wine to food choice. It is the intent behind the search that leads searchers to be satisfied or unsatisfied with the results returned.

This is the reason why 53% of all searchers use up to three separate engines. We keep searching until we get the "right" results.

{ 1 comment }

I was intrigued by a patent that was awarded to Yahoo this week. The patent mentions auction sites in regards to search engine results and relevance.

From paragraph eight: “Matching outdated, e.g, eBay listings pages to search queries can erode search user trust. The relevance problem that occurs when a query matches the title of an expired listing contributes to this.”

Patents are often written a vague as possible to hide the true intention of the idea/technology they are trying to patent. This patent seems to revolve around auction listings, referencing them no less than 10 times and specifically referencing eBay twice. It would make one wonder why Yahoo is focused on auctions. Are they concerned with delivering more product related returns?

Nope.

Obviously search result that goes to an outdated eBay auction would erode confidence. Not in that a user would care so much that the auction is over, but in that the search results were not fresh.

And that is what this patent is about – freshness. Auction sites provide a good example of the problem they are trying to solve yet is oblique enough to the true objective – indexing more sites much faster to provide very fresh results.

Why mask this? Online forums have seen their share of webmasters complaining about the amount of bandwidth search engines spiders consume – which is a problem since most hosting companies charge on bandwidth consumption. Jakob Nielsen covered this topic in the past.

Yahoo doesn’t want to bite the hands that feed them (webmasters) while trying to serve the freshest food (sites) to their consumers (searchers). I applaud this on many levels. As a search marketer, I want my client’s sites to be indexed as often as possible, and as an active searcher I want results that are up-to-the-minute fresh.

Yahoo is trying to find ways to push the gas pedal down while trying to respect bandwidth.

From paragraph 22: “This enables the search system to use update cycles which are best suited for particular types of content or listings, as well as reduce the number of queries required to maintain fresh and relevant content.”

In other words, update your site often and Yahoo is more likely to index more often. Yet they are clearly working on a system to reduce potential server load.

Technorati Tags: , ,

{ 0 comments }

Baseline has an excellent primer on Google. It’s lengthy, but a great read if you want to see how the giant works.

Link: Baseline: How Google Works

Technorati Tags: , ,

{ 1 comment }