November 29, 2005

Cybermundanity

How was your Cyber Monday? In case you missed out on the avalanche of hype, online retailers promoted yesterday as "Cyber Monday," a brand-new coinage. While "Black Friday" — the day after Thanksgiving — kicks off the traditional holiday shopping season, online shopping is supposed to follow suit on the following Monday with a big spike in purchases. But the Monday after Thanksgiving has in the past proven to be only the the twelfth-biggest online shopping day of the year, according to Business Week, which exposes the new name as little more than a clever marketing ploy:

It turns out that Shop.org, an association for retailers that sell online, dreamed up the term just days before putting out a Nov. 21 press release touting Cyber Monday as "one of the biggest online shopping days of the year."

The idea was born when a few people at the organization were brainstorming about how to promote online shopping, says Shop.org Executive Director Scott Silverman, who answered his phone, "Happy Cyber Monday." They quickly discarded suggestions such as Black Monday (too much like Black Friday), Blue Monday (not very cheery), and Green Monday (too environmentalist), and settled on Cyber Monday. "It's not the biggest day," Silverman concedes. "But it was an opportunity to create some consumer excitement."

Still, as far as marketing ploys go, the coinage of "Cyber Monday" has by all accounts been an overwhelming success. (The neologism-hunters have been hot on its trail: Paul McFedries of The Word Spy reported it on Nov. 25, Barry Popik of the American Dialect Society on Nov. 26, and Grant Barrett of Double-Tongued Word Wrester on Nov. 27.) The first known media mention of "Cyber Monday" was in a Nov. 19 New York Times article, as the Times evidently had the inside line on the machinations at Shop.org. After the official press release two days later, "Cyber Monday" spread extremely quickly, aided by numerous mentions from both traditional and online media outlets.

And how might we measure the initial spread of the coinage? Not surprisingly, the Business Week writer takes the easy way out and trumpets a Googlecount in the article's lead paragraph:

Do a Google search on "Cyber Monday," and you get as many as 779,000 results. Not a bad haul for a term that was created just a week and a half ago to describe the jump in online shopping activity following the U.S. Thanksgiving holiday.

A count of 779,000 is impressive to be sure (though the article is careful to hedge its bets by saying "as many as 779,000 results"). But as we've seen time and again, Googlecounts of any sizable magnitude simply aren't trustworthy. There's no reason to expect this number to be particularly meaningful.

At the moment, Google tells me there are about 740,000 results for "Cyber Monday." No matter: Googlecounts can vary quite a bit according to time and place accessed, and that's only 5 percent off from the Business Week count. I then cross-checked the count against Yahoo, which search-engine observers such as Jean Véronis find to be generally more consistent in its reporting of counts returned (though it too has its shortcomings). Yahoo currently yields about 751,000 results for "Cyber Monday," so there's no big discrepancy between the two search engines.

If we skim through the results on both search engines, however, we'll see a lot of repeated text. Theoretically, this should be offset by the search strategy I previously discussed, where one finds the count for "most relevant results" by appending "&start=950" to the end of the URL (for Yahoo, "&b=1000" does the trick). But this yields some odd conclusions. Though both search engines max out their "most relevant results" at about 1,000, only Yahoo reaches that limit for "Cyber Monday" — Google reports only about 500 distinct results!

Surely there must be more than 500 appearances of the term that Google considers non-identical. I'd expect this to be yet another spurious figure, based on some quirk in the algorithm Google uses to compare pages to judge their similarity. Even if this number is far too low, the repetitions of text involving "Cyber Monday" are extremely frequent. For instance, many sites reprint a Nov. 25 article by Reuters under the headline, "Online retailers await 'Cyber Monday.'" How many of the Google and Yahoo results use this text? Quite a large proportion, if the total results have even a modest degree of reliability. Here is what I currently get (though of course I cannot guarantee that these figures will come close to approximating anyone else's results):


Google Yahoo
"Cyber Monday"
740,000
751,000
"await Cyber Monday" 402,000
284,000
percentage using "await"
54
38

Disjunctive queries to determine the non-Reuters appearances are reasonably consistent with these numbers: <"Cyber Monday" -"await Cyber Monday"> at the moment yields 381,000 results on Google and 478,000 on Yahoo, roughly what one would expect by subtracting the above figures. Could a third to a half of the search results for "Cyber Monday" really reproduce the Reuters article, or at least its headline?

A look at some of the results for "await Cyber Monday" indicates that a huge number of websites include automatic feeds of news stories from Reuters and other wire services, mostly via RSS services provided by such portals as Yahoo News or CNET News.com. The Reuters headline was included in the feeds for top news stories, and it thus appeared automatically on countless sites.

Beyond the reproduction of the Reuters headline, it's also important to consider the provenance of this particular term. "Cyber Monday" was invented by online retailers for the express purpose of boosting their Web-based sales. It wouldn't be at all surprising if various websites were flooding the search engines with text on "Cyber Monday" in the lead-up to the big day as a way of building the hype.

This is yet another object lesson in the unreliability of counts provided by Google and other search engines: words and phrases thrust suddenly into circulation may show very misleading counts, at least until an initial period of volatility passes. (Further evidence for this volatility: I waited a few hours and checked Google again, and now it gives me 1,010,000 results for "Cyber Monday" and 585,000 for "await Cyber Monday"! I would expect the number of raw hits to continue fluctuating wildly over the next few days. Still, the "await" ratio doesn't seem to have changed much, with 58 percent of Google's results now apparently derived from the Reuters headline — not that I put much stock in that figure either.)

When the retailers count up their profits, they'll find out the extent to which the "Cyber Monday" hype paid off. Whether the marketing campaign had any lasting effect on the lexicon remains to be seen. Many of the results returned by the search engines (particularly those generated by news feeds) will soon fade away, and "Cyber Monday" may quickly be forgotten — at least until next year, when the inevitable marketing machinery kicks into gear once again.

Posted by Benjamin Zimmer at November 29, 2005 06:46 AM