You can debate the existence of an “advertising backlash” among consumers, but there’s no arguing with certain facts. According to Forrester Research analyst Chris Charron, 60 million households have signed up for the Do Not Call registry, 54% of households use software to block spam and pop-ups on their computers, and digital video recorders are on average permitting their users to screen out about 60% of TV ads in the programs they see. That may not constitute a groundswell, but at the least it’s a substantial minority of media consumers who have chosen to go on an advertising diet.
How do marketers respond to this consumer reaction? One way is to target their ads carefully, to make sure that the people who view their messages find them relevant enough to respond or at least see them. Online, that’s been part of the appeal of search marketing; linking to search keywords gives your ads a better chance of linking to the searcher’s current interests. And beyond the search page, contextual and behavioral advertising have risen up to infer consumer interest, either by what they’re looking at the moment the ad is delivered or by what they’re viewed on the Web in the recent past.
Web start-up mSpoke is taking a very Web 2.0 approach to targeting, by allowing web viewers to rate not only how closely the content they’re looking at relates to their interests but also how well the online ads they’re viewing have figured out who they are and what they’re interested in.
It’s a new slant on an old Internet idea, says marketing director Matt Fleckenstein. Web personalization has been around since the early ‘90s, when it was mostly folded into CRM tools, he says. Later in the decade, a second wave of companies built on that work and became the basis for a lot of the e-commerce recommendation engines that are out now, used by companies such as Amazon.
Today, consumer-generated content looms larger than ever on the Web; blogs tracked by Technorati have doubled in the last year to more than 33 million, and YouTube and MySpace have opened floodgates for consumer content. “That’s driven a need for ways for consumers to sift through all the stuff that’s online and find things that are truly relevant for them,” Fleckenstein says.
And that drive to personalization also creates opportunities for Web advertisers and publishers, he says—not to mention for next-generation personalization solutions like mSpoke. “We combine technologies like machine learning and consumer input to provide what we claim is the most personally relevant advertising and content on the Web,” Fleckenstein says.
Here’s how it works for consumers: When they go to the Web site of a content publisher that’s using the mSpoke platform, they will be asked if they want to create an mSpoke profile. If so, they enter their birthday, ZIP code and a few keywords that outline their primary interests. Within a few moments, the Web page and its sections will re-shape to fit those expressed preferences.
Once that base information is in the system, the mSpoke platform goes out looking for more articles and ads that relate to those interests. Consumers can take charge of what they see in two ways. They can alter or refine their preferences directly in their profile, telling the mSpoke algorithm, for example, that they always want to see stories about bicycling but only usually want to see soccer items.
Alternatively, they can pass judgment directly on the items the mSpoke platform has served up by clicking on little “thumbs up/ thumbs down” icons next to the content. (It’s very cool; on the My PittsburghLIVE.com page, thumbs-down articles fade slowly to a wispy gray before your eyes.) The mSpoke algorithm will then apply adaptive machine learning and, next time the consumer logs in, will favor the type of content that’s drawn ayes over the stuff that’s earned nays.
The engine is also able to learn implicitly just by looking at what stories the consumer clicks on to read further and what gets ignored. In ad terms, if an ad matches a keyword a consumer has in the profile, it will get served. If it matches the context of an article for which the consumer has expressed a preference, it will get served. And finally, if a consumer has expressed an ad’s relevance behaviorally—in other words, by clicking on other ads from the merchant or on the subject in the last four days—that ad will also get delivered.
One flaw in both personalization schemes of the past and the contextual and behavioral targeting systems used today, says Fleckenstein, is a low level of consumer transparency. “They haven’t really involved consumers in the process,” he says. “Users need to understand why they’re being shown certain content and certain ads, and—down the road—why they’re being recommended certain products.
MSpoke users can click on an icon to learn what about the rules the algorithm has assembled for them has led to their seeing a particular item or ad. They then can refine or discard that rule to suit their preferences, and the algorithm will use that new knowledge in future choices.
Linking both online content and online ads to consumer control is unique to the mSpoke platform, Fleckenstein says. “We’re the only company we’ve come across so far that’s really focused on personalizing both content and ads,” he says. “For a long time, those two areas have been treated as completely separate. But at the end of the day, it’s all part of the user experience, and there are a lot of synergies between the two.” Serving up content and ads on the mSpoke platform can make both more engaging for readers.
That’s good for Web publishers, because it forges deeper bonds with their audience and gets them coming back more often and using the site more extensively. It’s also good for publishers because it’s good for advertisers, Fleckenstein says. More personally relevant ads means more highly qualified clicks and better conversions—and thus more revenue for the publishers.
“It’s important to understand that we’re not giving consumers the Tivo-like power to skip past ads online,” he stresses. “Consumers have come to appreciate that if they want to get free high-quality content, they have to give publishers in exchange the ability to serve up ads.” But that doesn’t mean that they can’t have some power—in fact, ultimate power—over the relevance of those ads.
The mSpoke platform will work for the same reasons that behavioral targeting works, Fleckenstein says. “Because you’re targeting your message to consumers who are in the mindset to absorb it, you’ll achieve greater campaign lift. You’re not wasting impressions on people who aren’t interested.”
The Pittsburgh-based company launched in March of this year after spending months in stealth development mode. Co-founded by CEO Dave Mawhinney and CTO Dean Thompson, mSpoke is building its market to crucial mass by licensing its platform out to both traditional Web publishers and to smaller new media sites and bogs. The Tribune-Review Publishing Co., owner of the Pittsburgh Tribune-Review, was the first publishing customer signed, and mSpoke is currently in talks with CMP to put a personalized version of InformationWeek.com on the platform.
“As an entry point into the market, we’ve chosen to go out and license our adaptive personalization engine to traditional online publishers,” Fleckenstein says. “But we’re also in talks with Web portals and new media publishers. That’s a way to get lots of consumers using our new technology, but in time we see this platform being relevant not just to text content but to other types of online media—photos, video recommendations, podcasts and videocasts, music, movies.
“Eventually, we envision that consumers will be able to move around sites that use our technology and have the adaptive engine serve up the most relevant content, ads and even products to them using what we’ve learned about them from other sites.”