If there’s one way to sum up the resurgence of display advertising compared to search advertising in the last few years, it would be this: Display is finally a market. Like a financial security that can be split, swapped, sliced, diced, packaged, unpackaged, and hedged, ad space is being matched to the exact right bidder, at the exact right price. This is putting a floor on the price of the worst inventory, and pushing better ad inventory towards pricing that more accurately reflects its value.
And none of this would be possible without numerous layers between advertisers and the sites on which they advertise. The infamous Luma Partners display ad infographic makes this point nicely: there are alot of tiny optimizations that determine which advertiser wins which inventory, and every single layer appears to be adding value.
But there’s a built-in risk to this entire system: every transaction has quantitative and qualitative guidelines, and while quantitative guidelines are easy to maintain, qualitative guidelines tend to get diluted at every stage. GM cares deeply about what kinds of sites run their ads. GM’s agency cares, too. The exchange GM’s agency uses to buy its ads probably doesn’t care too much about GM per se, but does care about its relationship with the agency. The ad network cares a little less than that, and the publisher is mostly happy for the revenue.
Which is why GM is so understandably upset that their ads are running next to porn. This is the almost inevitable result when someone drives hard bargain in terms of quantitative limits (i.e. buying a certain number of pageviews at a certain CPM), but can’t clearly delineate the qualitative guidelines (what’s a trustworthy site? What’s a classy site? What’s the boundary between a celebrity showing skin and a site showing softcore porn?). GM could whitelist some kinds of content—but if they created a whitelist, surely Yahoo-owned Flickr would be on it. And if they’re already picking where they want their ads displayed, how much do they really need outside help in the first place.
This is clearly demonstrated in this excellent Digiday piece on how a single advertiser with a large budget and a focus on CPMs was able to alter best practices in the video ad market to the point that quality was seriously threatened.
At no stage in these transactions do people need to think that they’re being dishonest. It’s not necessary; there are not sharp delineations between legitimate activity and running ads that hurt the company that pays for them. There’s just a drift as the transaction that places an ad gets further and further removed from the business interest that conceived the ad in the first place.
In some ways, this is a classic agency problem; another variant on the problem Groupon ran into with its superbowl ads. But in another sense, the fungibility of online ads makes them even harder to handle than traditional ads or other agency relationships. A few possibilities:
- As they’re already doing, larger companies may need to focus on trusted, non user-generated sites.
- There’s now greater economic demand for algorithms that can detect offensive content. Search engines surface offensive content comparatively rarely; perhaps DoubleClick will be the first ad network to offer an effective automated filter, since they can lean on Google’s search engineers and their existing corpus of data.
- There might be demand for an independent accreditation service, able to rank sites by some standard (e.g. sites that are an “A” for coastal twentysomethings with masters degrees may be a “C” for middle-aged midwestern evangelicals).
This isn’t catastrophic yet. It’s likely that more people noticed the incongruous ads on Flickr through the press coverage they got than through browsing Flickr. But few CMOs want to face even a tiny risk that they’ll run their ads against offensive content. Right now, the economics of display advertising make it very expensive or very dicey to avoid.
(This is probably why Facebook only recently launched their ad API. Facebook has focused on brands for strategic reasons: if they can own branded advertising online while Google gets direct response, Facebook will be a far bigger business. But this means they have more to lose from making brands look bad. A related problem: some of the highest-CPM ads don’t fit into the normal ad exchange format.)