• by Laurie Sullivan  @lauriesullivan, December 9, 2010

Marketers can gain a true understanding of campaigns through randomized search experiments. So says Jim Koehler, senior technical lead for search ads at Google, who on Thursday during the opening keynote at the Search Insider Summit in Park City, Utah, put a framework in place to help marketers understanding the importance of experimentation when creating a successful search campaign.

Koehler addressed managing spend with bids or budgets, bidding to maximize profit, randomizing geo-locations, and determining the media mix. In nearly all cases marketers should manage spend via bids, he says. Marketers running budget capped campaigns, where no additional funds are available, lowering bids could prompt more clicks.

As an example, he illustrated how Google took a national campaign and modified half the queries, allocating lower bids to 25%, and then uncapping the cost for the remainder to examine the variations in performance. In the end for an uncapped campaign, it cost the brand 164% increase, but they got 166% more clicks.

Overall for the lower-bid queries, the spend was lowered about 37%, but the campaign still received 20% more clicks. Koehler's advice, "use the bids as the mechanism to change the spend." Once that's done, marketers should adjust the bids and the budget spent, so the marginal cost per clicks (CPCs) are equal to the value of the click. Of course, the results from the examples are "just examples and marketers should do their own experimentation to find their own results," Koehler says

Marketers also should measure the influence of online campaigns on in-store sales. During this experiment, the goal to measure an increase of in-store sales from the online campaign resulted in a short delay in offline sales as the campaign ramped up. When the advertiser turned off the online campaign, offline sales continued for weeks later.

During this six week test of 10,000 stores and 5,000 ZIP codes, the test showed that for every dollar spent in online advertising they would get about eight times return on in-store sales. While the campaign drove traffic to the company's Web site, sales didn't occur online, but rather offline.

Ironically, there was no lift to online sales. Clearly the online campaign drove traffic to this manufacture's Web site, but the manufacture did not gain incremental sales from the paid search campaign.