Google on Improving Adsense/Adwords

What are the best ways to pay someone for displaying ads on their sites? What are the easiest ways for people to find sites that they may want to have their ads placed upon? What information should be shared with advertisers about the sites that they might want to advertise upon, or have chosen to place ads on?

Adsense and Adwords are two sides of a content-based advertising system used by Google, and are amongst the methods that the search engine relies upon to make money. One of the main issues that faces Google is finding ways to make it easy to match up Adwords advertisers with the sites of people who display Adsense ads.

A new patent application from the Mountain View based search engine describes a method to help people looking to place ads with sites that are rich in content, have a lot of traffic, and are good prospective advertising hosts.

The patent filing is Determining prospective advertising hosts using data such as crawled documents and document access statistics (US Patent Application 20060095322), and lists Timothy Matthew Dierks as its inventor. It was originally filed on November 3, 2004, and published on May 4, 2006, and appears in the USTPO assignment database as being assigned to Google.

Here’s the abstract:

Ad delivery systems want to find good advertising partners easily and efficiently. To this end, available data such as crawled Webpages, access statistics, advertising offers, etc. may be analyzed. The available Webpages may be scored and sorted based on estimated revenue of the Webpages. The scored and sorted Webpages may then be filtered to remove documents considered to be poor prospects and/or documents having characteristics that are considered to make the documents poor prospects, and then presented to the ad delivery system for further use.

Summary of Method

A very basic summary, this method may be used to:

  1. Accept webpages as advertising sites,
  2. Score the pages in terms of things like:
    • Expected page views,
    • Expected ad revenue per page view, and/or
    • A product of expected page views and expected ad revenue per page view,


  3. Sort the documents using their scores, and present them to advertisers.

Some of the documents could be filtered, if they aren’t likely to be good prospective advertising partners, such as government web pages, or other pages that are known to have a policy of excluding advertisements.

Information provided to Advertisers

Here’s some of the kind of information that might be given to an Adwords user, about an Adsense site, to let them decide if the site is a good match for their ads, and to let them know about how their advertisement is doing once upon the page:

1) Features of, and information on, a page which an ad might be served upon, including one or more topics or concepts determined to be associated with the page,

2) Information or content located on or within the page,

3) Information about the page such as the host of the page (e.g. AOL, Yahoo, etc.),

4) The importance of the page as measured by things such as traffic, freshness, quantity and quality of links to or from the page, etc.,

5) The location of the page within a directory structure,

6) A search query or search results associated with the serving of the ad,

7) Characteristics about the users of the page, such as their geographic location, language used, type of browser used, previous page views, previous behavior, etc.,

8) A host or affiliate site (e.g., America Online, Google, Yahoo) that initiated the request that the ad is served in response to,

9) An absolute position of the ad on the page on which it is served,

10) A position (spatial or temporal) of the ad relative to other ads served,

11) An absolute size of the ad,

12) A size of the ad relative to other ads,

13) A color of the ad,

14) A number of other ads served,

15) Types of other ads served,

16) Time of day the ad is served,

17) Time of week served,

18) Time of year served, etc. ,

19) Other serving parameters.

Valuing the Economic Value of Ads

Page views are one important consideration when scoring the value of an ad on a specific site, and the document discusses how information about traffic to pages might be gathered from sources such as toolbars or by being purchased from internet service providers.

In addition to estimations of value on the basis of page views for a target site, other factors are taken into a consideration to determine an economic value, such as an analysis of the content of the web page to identify ads that would be relevant to viewers of the page, and an estimation of the economic value of displaying such relevant ads, based in part by whether the ads are cost-per-click offers, cost-per-impression offers, etc.

Other factors include the amount of potential available ad spots on the web page, the topic or topics of the web page, and information about ads targeted to the topic.


This document describes a process that might have a lot of value to people who want to show ads on sites through Adwords by making it much easier to find sites that are good matches for their ads. It also provides some ideas that people who make money displaying ads might want to think about when putting together a site, or deciding where to place advertisements on their pages.

If a system like this went into place, I would be pretty curious to see the score that Google might assign my sites.