Performics, the search engine marketing division of DoubleClick, came out with another flight of its quarterly Performics 50 keyword performance index in mid-February, and this report, covering the last quarter of 2005, sheds some insight on what happens to search terms during the holiday sales season.
For one thing, campaigns are getting larger and attracting more clicks, according to the “Search Trend Report Q4 2005”. The number of active keywords managed in campaigns between October and December of last year increased 58% over the number managed in Q4 2004. The sequential increase was much smaller—only 8% from Q3 2005.
But clicks on those keywords grew 48% from Q3 2005 to Q4, and that jump was eclipsed by a year-over-year increase of a whopping 108%. (The Performics index only includes active keywords.)
Some of these increases can be explained by greater commitment on advertisers’ part to expand their search marketing efforts, says Performics director of search strategy Cam Balzer. “Campaigns are getting larger, and advertisers are making sure that they’re there for that search demand,” he says.
Balzer says “Every Q4 we’re prepared for huge growth, but when it comes, you just can’t believe the volume,” he says. “It wasn’t so much the percentage magnitude of the growth but the sheer volume, especially in light of concerns about how the holiday season would shape up, with concerns over natural disasters and such.
The Performics 50 index measures not just the cost of bidding on a keyword but also the cost of owning that keyword; in other words, it combines both the bid price and the price of the clicks that keyword then evokes into a measure called cost per keyword (CPK). And since clicks grew more than the number of keywords in the quarter, it makes sense to expect that the average CPK of the terms used in the Performics 50 index increased.
Indeed it did, rising from $24.50 at the end of Q3 2005 to $54.62 by the end of Q4. roken out monthly for the quarter, CPK increased 75% in October, 100% in November and 57% in December. That’s not quite as steep as the climb during the same quarter last year.
“If you look at it month over month, the increase in CPK this year was actually lower than it was last year,” Balzer says. “That may reflect a more gradual tapering, or advertisers tempering their aggressiveness slightly and learning from the previous holiday season.”
This past holiday season saw increases in cost per click (CPC), a trend that acted as an accelerator on the cost of managing a keyword. Part of that quarterly CPC increase can be put down to the holiday effect—with more consumers using search and querying more keywords, advertisers will be willing to pay more to reach those searchers—but Performics has identified another driver behind higher CPC prices: A reduction in the use of lower-priced terms among the companies in the Performics 50.
According to the report, the number of keywords priced at 10 cents a click or less has been dropping since about July, when they made up about 40% of keywords in the index. These low-CPC keywords now comprise about 25% of the index. The timing suggests that the falloff is related to Google’s mid-2005 introduction of quality-based bidding, which links minimum bid price not only to other bids for a keyword but also to the “quality” of an ad—that is, to the number of clicks it commands. At least on Google, low-cost keywords may be undergoing a price hike due to low-clicked ads.
“Keywords that are on average lower-priced are getting more expensive,” Balzer says. “And it looks like they’re dong that at a higher rate than the overall average is going up. Google used to have a flat minimum fee of 10 cents, and a lot of keywords could get volume at that 10-cent minimum bid. Now they have a sliding scale, where a given keyword may have a minimum bid of a penny but others may now have a minimum bid of 50 cents.
The Q4 2005 report also found that, both this holiday season and last, higher-ranked keywords performed better in November than in December. For the last month of the year, ads on lower-ranked keywords have a bigger impact on conversions than high-ranking ones. In December 2005, users were more likely to convert from search ads that appeared lower than second position for a keyword. It’s an effect that Performics also noted in the 2004 holiday season, although this past year it was less pronounced.
“It may be that as consumers adopt search more and more as part of their holiday shopping activity, they’re being more thorough, looking further down in the sponsored listing and clicking on lower links,” Balzer suggests. “This really emphasizes the need for advertisers to be represented in all types of search during this period, even if they can’t afford a top listing.”
Balzer suggests that marketers can use the data from the Performics 50 as a benchmark to evaluate their own campaigns. “They can use the same kind of methodology we’re using—the cost per keyword—and benchmark their own metrics against this subset of representative managed campaigns,” he says. “If their own growth is at pace with this market, that gives them a good indication that their campaigns are performing well.”