ContextWeb Steps Into the Contextual Advertising Ring

ContextWeb stepped into a crowded ring of contextual advertising solutions today with the launch of ContextAd, and news of $3 million in venture funding.

Like Google’s AdSense and Overture’s ContentMatch, ContextAd analyzes keywords and page content to match a page with relevant ads. Unlike its larger competitors, it uses a taxonomy-based algorithm to clarify the difference between identical words used in different contexts.

“We use real-time indexing that can put the words in context and tell the difference between Paris, France; Paris, Texas; and Paris Hilton,” said Jay Sears, VP business development and publisher relations. “We create a keyword hierarchy, so you’re not just working from one keyword, but 10 or 20. For publishers, we’re taking inventory that may not be identified and making it more valuable.”

In Sears’ example, one advertiser could buy the keyword “Paris” in the travel category, while another buys the same word in entertainment.

Anand Subramanian, ContextWeb founder and CEO, says the product takes the best features of keyword and category analysis and combines them to achieve a higher degree of relevancy.

“Search was about keywords, and people who come to contextual advertising from search want it to be about keywords. People who come to it from the content side want it to be about categories. This is a whole different business,” he said.

ContextAd assigns content on a page to one of over 500 categories, with three levels of depth. Advertisers bid for placement in a general category, or pick specific keywords in addition to that category. The process is intended to enable reach and targeting opportunities. Different creative executions can be tailored to specific keywords or to a more general category.

Contextual advertising startup Kanoodle takes a similar category-based approach to contextual advertising, delivering sponsored links via content targeting, rather than keyword matching. Their product uses static categories for each page, determined either by Kanoodle or the publisher. In contrast ContextAd analyzes a page’s content in real time.

“You have to be able to process in real time to serve advertisements that are going to be effective,” Sears said. “We pass the URL from the publisher and serve the ads in real time. If you’re working with ad networks, it’s possible that 30 to 40 percent of page calls are unique — URLs you haven’t seen before.”

“Because we’re doing this in real time, we can get scale through the ad networks,” added Subramanian. “Ad networks have new URLs all the time. If the targeting is not being done in real time, it’s useless.”

This real-time page analysis responds to changes in a site’s content by delivering new ads. The company hopes this will attract publishers whose sites have dynamically changing content. Often, page context changes between when it was spidered and when content — and the ad — were delivered.

ContextAd applies a business layer to the matching process, making sure the publisher fills available space not only with the most relevant ad, but the most profitable one. If a page with music content was determined a good fit for both a lower-cost ad about lyrics and a higher-cost ad about digital music, the higher-cost ad is displayed.

Advertisers have the option of buying and managing keyword “buckets” rather than individual keywords. The buyer chooses a category and a keyword, sees the minimum bid, and sets a maximum. The system’s auto-bid feature, similar to the one used by eBay, automatically increases the bid until the maximum is reached.

“It makes it more predictable. Each click is not equal, so it can’t be priced equally. We let supply and demand talk to each other. All we do is manage the process. Our aim is to make it easy, to ‘set it and forget it,'” Subramanian said of the bid process.

Initially, ContextWeb will utilize existing distribution relationships with MaxOnline, ValueClick and DoubleClick. The company has begun building its own network of sites, with initial focus on the finance, travel, high-tech and health and fitness categories. While these are mostly consumer-oriented, the company will consider building a B2B network if demand develops, Subramanian said.

In another possible application, ContextAd could be overlaid on a behavioral targeting system.

“Behavioral is about the user, and contextual is about content. Behavioral needs to scale to the entire Internet for a user to cross the network enough times to accurately determine their behavior. Contextual scales much faster. If you mix them together, it’s a much more powerful proposition,” Subramanian said.

While the contextual advertising space is crowded, there may be room for ContextWeb, according to Janice McCallum, senior analyst at Shore Communications.

” I think there’s room for someone with a targeted approach. Remember, nobody thought there was room for another search engine a few years ago,” she noted.

“I like the fact they’re using both the text analysis and a topical approach. It’s been either/or with most contextual ad firms until this point,” she said. “Their approach really does take care of the problem of taking words out of context when it’s just an analysis of the text.”

ContextWeb also announced a $3 million first round of venture funding, led by Draper Fisher Jurvetson (“DFJ”) Gotham Ventures, with additional funding by DFJ New England and DFJ California. The company also established a board of directors comprised of ContextWeb executives and representatives from DFJ and MaxOnline.