Databases Reference
In-Depth Information
The Advertiser's Bid for a Keyphrase
Advertiser
Ad Position
The Advertiser's CTR for Keyphrase
and ad combination
A set of advertisers
from 1 to n
Search Engine (who runs the auction)
Figure 8.3. Even better view of the sponsored-search auction.
a percentage of the search engine's traffic. This percentage of traffic has some corre-
lation to the ranked ad position on the SERP.
Most sponsored-search platforms use some type of quality score (i.e., an assess-
ment of the relevance of the ad copy and related content relative to a searcher's
query) for a keyphrase-advertisement combination, typically based on historical or
estimated click-through rate (CTR), as shown in
Figure 8.3
.
This quality score complicates the auction process for the bidders.
Generally, the search engine does not disclose the bids of the other advertisers (i.e.,
it is a closed auction). However, if the sponsored-search auction used just the bid, with
enough empirical data gathered by adjusting bids and seeing the resulting change in
ad placement, an advertiser could approach perfect information (i.e., determine the
bids of the other advertisers), assuming that aspects such as dayparting and person-
alization were not part of the picture (which they are
- and these further complicate
the auction). However, the introduction of the quality score nearly prevents this, as the
advertiser does not know the effect of the quality score on the ad's position.
The quality score algorithm is yet another aspect of the auction that only the
search engine knows.
If the quality score is held constant, then the advertisers could still approach the
kind of perfect information needed to bid optimally, even if the actual quality scores
are never learned. However, in competitive verticals, acquiring perfect information to
bid optimally, as defined by auction theorists [
10
,
6
], is difficult, as the quality scores
can continually change.
So, look at all the factors that are unknown in competitive market verticals.
The bids of the other advertisers may be changing from one auction to the next.
The quality score of the keyphrase-advertisement-landing page combination may be
changing as CTRs, ad copy, and landing pages change. New advertisers are con-
tinually entering or leaving the auction. Advertisers are constantly changing their
dayparting and geo-targeting options. Ads associated with a keyphrase via broad
matching options must be considered. There are just a lot of possible unknowns to
say with any certainty that an individual advertiser has perfect information.
However, these unknowns, in practice, are not much of a hindrance, as an adver-
tiser can get a reasonable bid based on reasonable information [
6
]. Moreover, the
number of unknowns makes it more difficult to game or spam the auction with vin-
dictive bidding [
10
]. So, generally, advertisers can get reasonable, albeit not perfect,
information.
