Microsoft Shows Off AdCenter

Microsoft last week invited its top advertisers to Redmond to dazzle them with the latest and greatest bits of its adCenter advertising platform and emerging projects from its adCenter Labs (adLabs) testing grounds.

A new release of adCenter is expected in the next 2 months, with many of the new features resulting from testing and customer feedback collected through adLabs. The update will contain some small usability fixes, and mark a pilot release of Microsoft Content Ads, the new contextually-targeted ad program that shows text ads on sites on Microsoft-owned sites.

While adCenter is in its early stages focusing on search and contextual text ads, future incarnations will include other kinds of advertising. Research projects in adLabs focus on seven key areas: keyword and content technologies, ad selection and relevance, audience intelligence, social networking, video, platforms, and devices. Projects are at various stages of development, with some nearly ready to be integrated with adCenter, and others existing as barely formed concepts.

Microsoft last week showed off adLabs research focusing on three key areas: keyword optimization, video display ads, and consumer orientation.

Keyword Optimization

Keyword optimization efforts are built upon the Keyword Service Platform (KSP), a set of services that third-party developers can use through APIs to build Web applications to analyze keywords. Microsoft released the APIs to developers last week, and Colborn said partners are currently working on various applications to more effectively suggest, categorize, monetize and forecast traffic and extract keyword terms.

“We’re making it an open network by publishing and sharing the algorithm,” said James Colborn, product manager for adCenter Labs. “We’re excited to see how our partners will build on the KSP.”

Microsoft has developed its own set of tools in adLabs based on the KSP, including a keyword forecasting tool, a search funnel analysis tool, and a search result clustering tool. A vertical competitive analysis feature allows advertises to dig deeper into the “long tail” keyword suggestions, and more accurately test budget predictions. Marketers will also be able to extract trends in search campaigns to create modeling and forecasting tools to apply to their broader business, since the search queries will capture user intent around the marketer’s products and services, he said.

Some of the more developed keyword optimization projects underway in adLabs include tools to improve the detection of commercial intent, to break down multi-word queries into separate phrases, and to classify keywords into a taxonomy that will improve contextual targeting of ads.

Commercial intent detection, currently being tested in adLabs, returns a probability score reflecting the likelihood that a word or phrase is part of a search with commercial intent. The scoring can be used to track a searcher’s progress through the purchase funnel, or to determine that a searcher is not looking to purchase, but to support a previous purchase.

For instance, “digital photography” shows a 51.4-percent likelihood of having commercial intent, which reflects the informational nature of the query. Moving farther toward commercial intent is “digital camera tips,” which shows an 81.7-percent likelihood.

When it gets down to choosing a camera model, searchers would likely turn to queries like “digital camera review,” which shows a 98.3-percent likelihood of commercial intent, and finally “digital camera store” shows a 99.8-percent likelihood of commercial intent.

“Most marketers are ROI-focused, and can’t afford the luxury of paying for non-commercial keywords,” Colborn said. The tool can also help marketers target ads to users that can help move them to the next stage of the purchase funnel, or target them with information or services appropriate to their non-commercial query. “Marketers understand that people do research and eventually buy. They may want to reach their audience in the research phase so they’ll keep them in mind later when they’re in the commercial phase,” he said.

In order to better process multi-word queries, Microsoft is developing ‘query entity detection” algorithms, to break down queries into chunks that go together. So a search for “real estate agent” would recognize the phrase “real estate” should stay together, instead of treating them as a collection of single terms. The algorithm analyzes the relationship between words in a search term, and decides the best way to break it down to improve relevance of results.

The keyword categorization engine will be incorporated into adCenter as part of the Content Ads addition later this quarter. It will help solve the problem that some contextual ad networks have of delivering too much irrelevant traffic or not enough traffic overall, Colborn said.

“You need to place contextual ads in the most precise category. The keyword categorization engine will identify content channels appropriate for a business, and provide a confidence score of which category a keyword fits in,” he said.

The Content Ads program has been in beta with several advertisers since October, and more U.S. advertisers will begin to see their campaigns show up in Microsoft’s Content Ads program as early as this week, according to the adCenter team. Advertisers were selected based on how well their ads match the inventory currently available in the pilot. Plans call for a roll-out to all customers shortly after the pilot concludes.

Unlike most other contextual ad programs, Content Ads is not yet open to third-party publisher sites. Instead, ads will appear on Microsoft properties, such as MSN, MSN Money, Dating and Personals. Advertisers will have the option to opt out of the Content Ads program at any time, or to set separate bids for keywords to display on content pages.

Many search marketers have seen early success with adCenter, though a common lament is for more traffic, as long as quality does not go down. Microsoft is hoping Content Ads will be one answer to the inventory problem.

Video Display Ads

Video ad research includes creation of tools to encourage social interaction around videos, such as enabling users to highlight and comment on a specific area of a video frame through Javascript and Windows Live ID authentication. An example would be allowing users to comment on a particular play in a sporting event, or a segment of a newscast.

Microsoft is far along on a video hyperlink product that’s set to debut in the spring with a large retailer. The tool lets marketers embed video hyperlinks within online video ads, which can be associated with specific products in the video. When viewers mouse over a highlighted product, links will appear that will lead to Web sites where they can find additional information or purchase products. Videos with embedded links will show a special icon in the lower right corner indicating their presence.

Consumer Orientation

Consumer orientation efforts involve building technology to help determine whether a Web page contains content that the general public might consider objectionable, such as pornography, crime or terrorism. This “sensitive content detection” is also expected to reduce the likelihood of inappropriate Content Ads being shown within the context of an article.

Ultimately, the goal of all these bells and whistles is to give advertisers the tools required to make better decisions, Colborn said.

Some customers are already finding that adCenter’s demographic targeting capabilities are leading to campaigns that convert better, and at a lower cost, he said. Such is the case with AG Interactive (AGI), the Internet and wireless business unit of American Greetings, whose paid search campaigns on adCenter are managed by Range Online Media.

Range has been using demographic targeting to market AGI’s online greetings and graphics to MSN’s audience since adCenter first launched its pilot in late 2005. Prior to testing demographic targeting in adCenter, AGI’s campaigns in adCenter delivered similar conversion rates and average cost-per-acquisition (CPA) as Google and Yahoo.

AGI’s analysis had led them to believe going into the test that its primary audience was made up of women over 25, with a core audience of women between 30 and 55 years old. In a test account, Range applied demographic targeting to active orders through incremental bidding for women 25-35, 35-50 and 50-65.

By doing this, Range was able to improve conversion rates on AGI’s campaigns by 3 percent overall, and reduce average CPA by 6 percent. At the same time, the targeting allowed AGI to target more active users to offer subscription services. Total subscriptions increased by 4 percent at a reduced cost of 2 percent, and the targeted demographic groups produced a 31-percent increase in subscriptions at the highest price level. The overall result was a 42-percent lift in incremental ROI.

Search Headlines

NOTE: Article links often change. In case of a bad link, use the publication’s search facility to search for the headline.

Related reading

Simple Share Buttons