Where Have I Seen That Scarf Before?
Yahoo News this week offered audiences a chance to sport the same scarf and pants as an individual who appeared to be involved in a murderous melee in Kabul last week. Now, it is entirely possible that this fashion plate was just in the neighborhood and crouched down to examine some of the spent shells involved in the killing spree and in fact is merely a trendsetter and not one of the "Taliban gunmen." But nevertheless, he was employed (shall we assume without consent?) as a marketing opportunity by Yahoo's in-image ad system.
"A man holding spent .50 caliber shells looks towards the Spozhmai Hotel on Qargha lake on the outskirts of Kabul on June 22, following an attack by Taliban militants. A carefree birthday party and night out to celebrate the start of the Afghan weekend were reduced to destruction and butchery when Taliban gunmen stormed a lakeside hotel, opening fire indiscriminately on wealthy revelers."
According to GigaOm/PaidContent, this is not the first time that in-image advertising has gone astray. It cites a 2010 NPR report that, when the technology first appeared, it offered readers a chance to “get the look” of a dying Elizabeth Edwards and a rehab-bound Lindsey Lohan.
Said one wag in the comments section: "I'm just impressed that they managed to find a matching scarf. Been looking for one like that for ages now. Well done Yahoo! Why do assume this is a marketing fail? Surely Yahoo needs all the publicity it can at the moment?"
Meanwhile, Facebook is paying out $10 million to an undisclosed charity in settlement of a lawsuit for using member names and profile pictures without their consent in ads. The plaintiffs argued they should be compensated for being used as unwilling spokespersons in Facebook's advertising efforts. I assume the fellow in the photo has retained counsel.
It does not take much looking to find lots of examples of inappropriate Internet advertising, where algorithms fail to deduce the context of a photo or story before deciding it is an appropriate match for an ad. The quandary is that online is the only media where this is a problem. While there is lots of stupid advertising out there (even TIME Magazine offers up a Top Ten list of Tasteless Ads) in the online industry we compound the problem by 1) deploying machines to make decisions they clearly can't always make correctly; and 2) trying to monetize every square inch of online real estate (something users hate like hell).
It's kind of mindboggling, the amount of talent and money thrown at new ways to help publishers monetize traffic. Many of them are new technologies, to split hairs that have already been split three or four times by other technologies. While the stated goal is to be more precise in making sure the right person sees the right ad at the right time, much of it is based on machine learning that will have to scale to billions of transactions before it has "learned" enough to be effective. And as we have seen, lots can go wrong in the process.
We seem to be drifting into the same fog that surrounds Wall Street trading, where only a penny or two is made at a time, but made again and again over billions of transactions. As the late Everett Dirksen apparently never said: "A billion here, a billion there; pretty soon, you're talking real money."
One wonders if on the march to get to those billions, we are creating a cesspool of marketing mistakes that will leave users more angry than amused.