Bambi Francisco, who’s one of the better, (yes, just my opinion), net watchers/bloggers/writers out there did an article on MySpace and talked breifly about their opportunity to target ads by category. However, the thing is, like so much else I’m personally seeing these days, a lot of these concepts are not very new. It’s just great that ‘net technologies may be catching up to early ideas such that these ideas can maybe be actually implemented. But – I think anyway – historical context is important. So here’s my thoughts…

The original article is here: MySpace-Engine Article (Bambi’s Blog)

From the original article: “If that’s the case, I’d argue that on the Web, global communities based on shared interests are actually the newest forms of clustering audiences for advertisers.”

There’s nothing new about it. It’s just never been done well. Geocities had the capability to target ads by community in the mid-90s, (albeit just banners, not text mapped contextually via keyword
matching). About.com certainly used and uses the ability to target communities of interest. These were early efforts, though they were hard to sell to advertisers. (Way back in the day, media budgets for online were still coming out of small r&d test budgets rather than media budgets as they are now. They didn’t see online as formidable, much less “get” the idea of context. Nor were there good contextual mataching tools and to this day they’re arguably not nearly what they
could be. E.g. the oftentimes click through difference between search keyword matched ads vs. broadmatch/site matached ads. Not to mention the lack of meta data for banners, which are essentially most often targeted by hand via ad network traffic tools.)

To be sure, there may be some value to ad targeting based on general categories in which pages happen to fall. It’s perhaps weaker context than keyword matching. However, that’s hard to say for sure since no one’s done clear enough research on the category level to definitively say just how much better. (At least not that I’ve seen.) How to do this will take some thought. The challenge in doing so in a MySpace type network is the same as it ever was for the seminal online services and
anything that’s out there now; the pages are generally not focused enough to do so. And even keyword contextual mapping will be questionable due to the disparate content on a page. Behavorial
targeting of ads may work a bit better in this environment. One benefit advertisers MAY have in such a network is potentially a higher click-though rate regardless of context. Why? Because often the content is so lame that there’s less user engagement and a higher propensity to click away. In fact, the point made above that the Hitwise data may indicate people jumping away to Google to continue to search may lend some admittedly small validity to this idea. (Plus, some old usage data I’d once had about an early personal web page publishing platform seemed to bear out the idea as well.)

So just for the sake of argument, say the communities of interest do have higher advertising value over a run of site. That means the pages would have to be attached to one or more taxonomic categories. No small trick. (Yes, I know. Various semantic search companies will claim to be
able to automate this. And this may be somewhat true. It’s still some effort to get right.)

Bottom line: Sheer volume may allow MySpace to generate piles of ad revenue. But as yet, there doesn’t seem to be much category-wise that would command a higher cost per thousand ad rate based on better context via category.