PR experts use a trick when they need to release news they really don’t want covered broadly, peer reviewed or scrutinized. The trick: drop the announcement when everyone is focused on other things. The Friday afternoon before a long weekend and the last business day before a major national holiday are prime dump days. The Bush White House used this tactic to announce Koran abuse at Gitmo and the indictment of Scooter Libby. Celebrities routinely use it to announce divorces or rehab stints.
And on December 23rd, just as the media world shut down for Christmas, Double Verify (DV) used it to announce its new “BrandShield” solution. Of particular note in DV’s release is that it seems to imply (the wording is quite cagey) that DV can perform page-level quality filtering on “nearly 100% of impressions”, even when ads are served within iframes, by effectively “seeing through” the iframes to determine “which…page the ad is actually delivered on”.
Taken at face value, this sounds like a huge advance in page-level quality filtering technology, which obviously requires page-level visibility to work. However, regular readers of this page will remember our recent post on the problems posed by iframes for 3rd party page-level filtering. Specifically, that “seeing through” iframes is impossible for an ad buy – like the vast majority of ad network buys – the composition of which is not known in advance.
So why would a (to date) publicity-hungry startup like DV announce seemingly ground-breaking technology in a way that recalls the indictment of a senior White House staffer? The only reason I can think of is that this announcement amounts to either a) an admission that DV is using the methods of hackers to exploit holes in browser security and enable collection of data that all commercial browsers prevent for important privacy reasons or b) a clumsy and misleading attempt to confuse the market about what is technically possible.
The former would raise extremely troubling privacy concerns, particularly against the backdrop of increased scrutiny on collection of user data for BT. The latter is obviously not particularly comforting either, but at least it doesn’t open unsuspecting agencies and brands up to PR backlash, consumer lawsuits and/or government sanctions. Either way, prospective DV clients considering this solution should ask tough, direct questions about how this apparent iframe miracle is performed before touching it with the proverbial ten foot pole. Specifically, buyers’ technical staffs should seek to understand clearly and precisely how each page in an ad buy would be conclusively identified and filtered, including each page where the ad is displayed within an iframe. As I mentioned above, be sure to consider the case where the composition of the buy is not known in advance, like most ad network buys.
Rest assured that we will be working with our agency partners to fully explore these claims and will share whatever facts we uncover on this page. Please feel free also to share with me anything you know or find out. As we set about that work (or at least until DV is good enough to clarify their release), I would renew my call for a New Year’s resolution: let’s elevate the dialog from misleading marketing claims to honest discussion and execution of the cutting edge solutions that sophisticated clients demand and deserve.
Andy's Blog 
Andy: You have quite the imagination. If you were really on your game you would have seen that the release was tied directly to a story that appeared in MediaWeek.com by Mike Shields. (http://www.mediaweek.com/mw/content_display
/esearch/e3idbd92851d6a66e352d8e0227975c9d91) and because it faced relatively little competition from other news his story got very nice pick up by other news aggregators. And so did the release. I think once BrandShield is released in mid-January, it will meet and even exceed your QA expectations for it. PS: I am DoubleVerify’s PR rep. Happy New Year (if that is not suspiciously too early for you).
Andy,
You have quite the imagination. If you did your research you would have seen that the release was tied directly to a story that appeared in MediaWeek.com by Mike Shields. (http://www.mediaweek.com/mw/content_display/esearch/e3idbd92851d6a66e352d8e0227975c9d91) , hence the timing of the release, and because it faced relatively little competition it got great coverage on many publications. I invite you to test out BrandShield and see the capabilities for yourself before jumping into conclusions.
Happy New Year.
Oren Netzer
CEO, DoubleVerify
Andy,
I think your posting is brilliant, and more complex then you can imagine.
Having had direct experience with DV’s solution, I am not shocked that they use the same tactics of the Bush Whitehouse. They have done a great job of scaring the market into thinking that the world is out to get the advertiser and they all should be scared and hide in their basement and not do anything without their survival kits (duct tape and plastic lining). I am surprised they have not implemented a color code chart yet on the safety of the web. Does that tactic sound familiar?
In fact, and I have data to prove this if you would like, the DV solution is a lot of smoke and mirrors. In my opinion, the data is immature (my analysis showed their classification had an error rate of double-digits), the technology solution is homegrown at best and their so called page grabs appear to be faulty, and they have annouced multiple products that don’t appear to actually exist. I believe any agency that falls for their scare tactics should be happy with their dwindling margins on their business for not having any originial thoughts and/or leading the market, rather than following it.
I applaud Brand.net for actaully calling these hucksters out. They are selling a bills of goods, and though George Simpson seems to be spinning quite a story, the rapid overreaction might tell you something.
Andy, maybe we can shed some light here, since at AdXpose we’ve been working on page level content and performance identification, and transparency, for 2+ years. In short, you’re right. 100% visibility is highly unlikely in the ad network and exchange world. We typically see a range between 60% and 80% of page level visibility on a typical client’s exchange based buy. The remaining impressions are dead-ended at either the ad server or network URL level, resulting (in the case of our service) in the ability to view aggregate engagement, performance, and quality data, but not content.
Now, this doesn’t mean services like ours cannot “see through” iframes. There are completely legal and ethical means (for which we have filed patents and which we share with our clients and prospects) of identifying referrer data and in many cases, we can see through two or three layers of iframes to identify which publishers or networks are “daisy chaining” an ad tag downstream via arbitrage. This is how we get to that ~80% number.
But when we cannot see below the server/network URL, we recommend to our most sensitive clients that they evaluate the level of confidence they place in that given network, and based on that, either ask the network for transparency, or simply cancel those portions of their buys. Otherwise, to your own recent point in iMedia, you can never be certain content adjacency issues will not occur, even on brand name sites.
In general (and we may differ here), we are big proponents of transparency. That’s how we differ most from Double Verify. We provide advertisers with the most comprehensive safety and performance data we can gather, and advise them on how to read it, but the decision on what to do with this visibility is up to them. We block ads from appearing on bad content *when we can see the content*, but we make no claims to do so when we cannot.
Trusting a third party to simply “audit” impressions or block ads from appearing without telling an advertiser where the offending placements appeared (when that very third party often makes its living by charging the network an audit fee) is not something we recommend – it’s the fox guarding the henhouse.
As an aside, our methodology *has* been vetted by third party research firm Radar Research, and we partner with comscore among others, for market-tested enhancements to our solution.
Great post and we applaud you for “elevating the dialog from misleading marketing claims to honest discussion” in the year 2010.