The words who, what, when, where and why are drummed into journalism students when they learn about how to construct news stories in inverted pyramid format. For example, “Paris Hilton (who) was bitten by her exotic monkey pet (what) last Wednesday (when) while both were shopping in a Paris lingerie store (where).” The “why” in this and many other stories is, of course, a lot harder to get. (Why on earth Paris Hilton would bring her monkey on a shopping trip is clearly outside the scope of this column.)
From a search marketing perspective, the “Five W’s” are especially important. While technology does not allow marketers to know exactly who is clicking on their ads or why exactly they clicked on it, we do know the what, when, and where, which allows us to make powerful inferences about the who and the why.
One important fact, which many new entrants to the search marketplace do not seem to realize, is the degree to which the success of their campaigns depends on accumulating enough historical data to make these and other inferences.
Some come armed with a lot of information about their customer’s historical behavior (including when they are in-market, where they are located and what types of messages they are most likely to respond to), and this information naturally helps them achieve better than average campaign results from the outset.
But others must “buy their way in” to this knowledge, and the only way to do so is to buy a bunch of keywords, throw up a campaign and let the numbers roll in. After enough clicks have occurred over a long enough period of time to comprise a statistically valid sample, the marketer can begin to make some informed decisions about what needs to be done to maximize conversions. Naturally, the more money that one has allocated to the exploratory stage, the quicker one can make these conclusions and begin the conversion optimization process.
For those marketers who must “buy their way in” to the optimization process, it might seem unfair that the search engines, each of which has a treasure trove of historical data pertaining to the way that keywords, campaign structures and dayparting strategies influence conversion behavior, do not share this intelligence with marketers buying keywords from them. Unfortunately, the degree to which the engines disclose such data isn’t likely to increase anytime soon.
In fact, all of the engines seem to be moving in the opposite direction (Yahoo, which still discloses actual, not predicted keyword prices, will no longer do so after its new hybrid-auction system goes into effect in 2007).
The fact that historical data has real, strategic value should not be lost on those who are using agencies to manage their search campaigns. Who owns this data the agency (which typically administers the account) or the marketer? What will happen to this data if the marketer decides to switch agencies (something which happens a lot in the SEM business)? These are all thorny issues that should be discussed upfront, before any deal is closed.
Historical information can give marketers excellent insight into the “Five Ws,” but its real power comes when this data is used with campaign automation system that makes use of this data to predict and model future keyword and offer performance and aggressively bids for those audience segments which are most likely to convert. In this scenario, historical data doesn’t just provide a path toward determining the Five Ws, but is actually the raw material from which future search profits are forged.
