In-Video Ads - How Does One Define Relevance?

Aug.23.07 | About: Alphabet Inc. (GOOG)

Scott Karp has a thoughtful post over on his excellent Publishing 2.0 blog about Google (NASDAQ:GOOG)'s roll out of in-video ads on YouTube.

The InVideo ad format is undoubtedly less annoying than pre-roll videos — and that’s a considerable achievement. But it’s nowhere near the kind of breakthrough that search advertising was.

Not being interruptive is the very LEAST that online advertising needs to do in order to thrive — what it really needs to do is be RELEVANT.

The beauty of search advertising is that the format and the relevancy of the ad are PERFECTLY aligned with that of the “editorial” content, through the miracle of search keywords.

That will surely be the case in some instances of InVideo ads, but in many if not most instances, the ads will have nothing to do with the editorial content — and the relevancy to any individual viewer, unlike keyword targeted search ads, will be hit or miss.

And there’s a BIG problem with low relevancy — advertisers only pay if someone views the ads.

I think the comparison to Ad Words and using the metric of contextual placement as in text ads might prove off base. It seems to me that much like contextual text ads superseded regular banners when Google crashed the party, a new form of advertising will supersede contextual ads in the video world. Current contextual ads, as Scott points out, are driven by keywords, or in other terms, the content on the page.

However, we need to remember that much of Youtube's growth was powered by MySpace and video embeds in personal pages and that is still where much video content is consumed. That itself implies a context beyond the content of the video. Moreover, that context is not understandable from within the video (this may be what Scott is getting at in paragraph 4 above). The context is a social one and relies on friending or recommendations of posters to their communities.

I think Scott is correct that contextual relevancy in video is much harder, i just wonder whether it is the context of the content as he seems to imply, or the context of the interaction that will determine that relevance.