An Ad App You'll Love (and Privacy Groups Will Hate)

NebuAd Tracks, Targets Consumers No Matter Where They Travel Online

By Published on .

A Push for Web Version of the Do-Not-Call List
NEW YORK ( -- A new ad network is about to step smack into the middle of a privacy maelstrom, thanks to a targeting system that could be good for advertisers -- but a tough sell to consumer advocates.

As the Federal Trade Commission met last week to discuss online-advertising targeting and its implications for consumer privacy, a potentially potent behavioral-targeting player was readying its launch. Today at the Ad:Tech conference in New York, NebuAd will unveil that targeting system, which uses internet-service-provider data to target ads to people based on where they are surfing on the web.

Most behavioral-targeting networks can record behavior data about a web user as long as that user is visiting sites within the network. In other words, its intelligence is only as smart as its network is big. NebuAd, on the other hand, partners with ISPs so it can observe consumers no matter where they travel across the web. It collects that data and uses it to serve targeted advertising based on where else the consumer has shown interest.

Interests vs. behavior
Sound Big Brother-ish? To be sure, NebuAd has prepared for a privacy outcry: The company presented its business to the FTC a month ago, said CEO Bob Dykes, a former executive at Juniper Networks. He said the company told the commission that the data it collects aren't identifiable and aren't stored and that all the advertising it intends to serve is of general commercial interest.

"We've designed it so we're only tracking people's commercial interest," Mr. Dykes said. "We're not keeping raw data on people's surfing behavior."

He said the company applies a "hash" to internet-protocol addresses so they cannot be identified when a user visits a site. The site is determined to be commercial or noncommercial (an auto site would be classified as commercial; a porn site wouldn't be commercial), and if it's the former, it's put into a category, and that category information is added to NebuAd's tracking data.

But such precautions haven't been enough to quell the concerns of some observers. Scott Bradner, technology security officer at Harvard University, wrote a column for Network World blasting the service. Later, after talking to Mr. Dykes, Mr. Brander said NebuAd was not as bad as originally thought, but he added that he's still skeptical of how the information could be used if the company "runs into financial difficulty or is bought by a less-scrupulous company."

"They've gone a little further than Google has but not in a quantitative step," said Mr. Bradner. "It's fundamentally the same thing -- it's tracking what you're doing. Google can't keep track of what you're doing if you're not talking to Google."

The company also has come under fire from tech-news sites TechCrunch and Slashdot, which both reported on an incident with a Texas ISP in which NebuAd was inserting ads on top of web pages without the publishers knowing about it.

According to a spokesman for NebuAd, the company serves ads only via agreements it has with ad networks. The company said business is built around an arbitrage model in which it buys inventory from ad networks, ideally at low prices. On top of that inventory, it layers the intelligence it has on web surfers and serves targeted advertising for which it can, because of its relevance, charge higher prices. It makes money on the difference.

NebuAds won't identify the ISPs it's working for, claiming that the service providers don't want to disclose their secret ad-revenue weapon. But it's more likely the ISPs want to tread lightly in an area that could be a privacy minefield.

One company specializing in performance, lead-generating advertising for clients including Procter & Gamble and the University of Phoenix has used the NebuAd network and said it performed better than other ad networks it has tried. Now, Focalex CEO Tom Soevyn said it needs to build out some scale. "I believe the model is pretty solid, but depends on how well they execute," he said. "Can they get the ISPs to sign on?"
In this article:
Most Popular