A recent Axios Media Trends newsletter contained a confusingly written piece that ultimately obscured an important point. While the headline dealt specifically with programmatic advertising (“Moffett: Programmatic advertising won’t save everyone”), the item delved into targeted advertising CPMs, as if programmatic is the only wayto sell targeted ads. While noting that AT&T claims that the merger will allow it to sell data-driven (i.e. highly targeted) ads against more video inventory, it went on to say
“…historically, the more precisely-targeted an ad is, the cheaper the rate (CPM) is, because you are reaching less people, and ad rates at the CPM (cost per 1000 impressions) basis are calculated based on reach, not effectiveness.”
A couple of things here. First, programmatic advertising is simply using computers and data to buy ad space automatically, frequently in real time. Yes, when it first came out, it was declared the savior of digital advertising. But, let’s be honest, it was marketers and publishers of small, low quality sites doing the declaring. High quality publishers saw it for what it was – a way for advertisers to cheaply buy targeted ad space across multiple sites regardless of the size of those sites. Size didn’t matter because if a likely BMW buyer was on a small blog or the Economist, the data trail identified her and the advertiser was happy to serve her an ad, on both sites. The problem was, while the blogger was now able to get advertising he previously couldn’t, a site like the Economist saw its CPMs fall.
But the CPMs didn’t fall because the ads were targeted, they fell because the ads were becoming a commodity. If the web is just one big content machine for marketers to place ads, even targeted ad inventory is infinite, and price drops. As bad as this was for publishers, it turned out to be too good to be true for marketers. As AdWeek noted:
“The programmatic open exchanges were discovered to be rife with all sorts of bad ad inventory issues. Threats include ad fraud, such as bots and domain spoofing, as well as brand safety issues (think your brand’s commercial next to a jihadi training video on some obscure website pretending to be The New York Times).”
So no, publishers, and now not even marketers, believe programmatic “will save everyone.”
Back to that confounding CPM statement. CPM, by definition, is not based on reach. It stands for cost per thousand, which explicitly removes the quantity from the number. And very targeted ads sold directly to brands by a publisher’s sales force DO tend do get higher CPMs. But they do have to be more targeted than simply geography and demographics. As we’ve seen, any data management platform can deliver that kind of targeting.
Later in the Axios piece Craig Moffett of MoffettNathanson says:
“And that’ll be, I think, one of the real controversies and challenges that the whole digital world faces, over the next decade — is will, in fact — better targeted advertising lead to more advertising revenue, or less advertising revenue? My suspicion is it will lead to less.”
Here, he makes the key point of the item. It doesn’t have to do with lower CPMs for targeted advertising (which he tends to suggest), but with the overall ad revenue earned by publishers selling highly targeted ad space. The problem, is that while your CPMs may be higher for that space, because it’s targeted and you therefore have a smaller quantity to sell, your overall revenue may likely be lower than selling your full inventory at a lower CPM.
That, is the targeted advertising problem.