Since Larry Page took over as Google CEO, there has been a renewed appreciation for his personality and influence, and how it affects Google’s approach to products.
Tip: if Larry wants something, he’s likely to get it. In the past, that often meant some cues being provided by former CEO Eric Schmidt, agreeing with Larry’s push. Of course, not to rule out Sergey Brin’s strong will in these matters either, but he’s no longer at the helm in any sense, so it will now be more Larry, more of the time.
Take Quality Score in the ad program, and the reasons behind it. If Google’s founders — based on clear signals sent from Google’s massive user base — didn’t care for the look of affiliates and other “yucky” advertisers, they would take steps to challenge them. They did this via rules at first, but later, most policies of this type simply got buried in an opaque, non-confrontational formula called Quality Score. (Keyword quality score was a continuation of the old Ad Rank formula, with additional sophistication; landing page and website quality score was a codification of policies, likes, and dislikes — driven mainly by what Google perceived to grossly offend some users.)
A top manager at Google once told us how unrelenting the founders were when it came to the perfection of policies and Quality Score algorithms; if the main objectives weren’t being reached in terms of real world quality and editorial filtering, they would keep hammering away until the team got it right. It was about perfecting the engine, not just setting up some kind of arbitrary scoring mechanism.
Indeed, the approach taken was not so different from that taken with organic search (which is a much harder problem given the openness of the web and the much larger database involved).
I noticed something recently that also — to use a term that has now become popular in the industry — “has Larry’s fingerprints all over it”. It’s in the conventional Display Network (formerly the content network).
Years ago, in the dark days of the unoptimized, fleece-o-licious content network, we routinely recommended that performance-oriented advertisers bid 75% lower, to equalize ROI with search. In other words, our belief was that the network would return about 25% of the ROI as paid search, so you’d have to bid that much lower to equalize ROI on each side.
As the network got smarter, that got to be closer to 50%, then 60%. For the past couple of years we’ve often recommended a bid guesstimate of around 75% of your median click cost on the search side in an ad group, again, in order to keep ROI even on both sides.
Lately, on certain kinds of high volume, relatively routine, relatively predictable accounts in the US, we’ve found the network to be increasingly uncanny in performing equally to search. That means, quite often, leaving the bid on display at somewhere around the same as your median click cost on search will give you about the same ROI. Granted, we have tools that help us bid even higher on some search network sources for some ad groups (managed placements), or to exclude or lower bids on the sources that don’t perform as well. But it’s amazing how this “roughly equal” result keeps recurring. More often than not, we’re raising bids on specific display placements about as often as we’re excluding or lowering them.
This has all the signs of a relentless quest by Google to achieve just that result. If it’s not Larry’s doing, then it’s “opinionated Googler X” that has pushed and pushed until this equilibrium has been reached. It’s good business. It creates confidence that Google’s full AdWords program performs for advertisers willing to tweak and optimize thoroughly.
What does it mean for you?
- Drop legacy assumptions about how the Display Network performs.
- Keep watching search like a hawk! Its foibles are now equally painful to the junky inventory you used to run into on the display side.
- Investors and analysts may need to be aware that Google thinks this way. It protects trust with advertisers and users, which means the cash cow that funds other stuff is at least secure.
- However, that other stuff is by no means proven to be anything but expensive research at this point. Google might grow, and there is limited downside risk in its core, but there is still a lot of risky behavior in other untried business models.
- Google will have to tell different stories to different advertisers. There is basically no way they can achieve a similar goal (rock solid performance) across all of YouTube and other less reliable areas of display advertising, so they’ll have the brand lift pitch for some folks, and the performance pitch — well, it won’t need to be pitched, as performance-based advertisers have been pretty much addicted to the performance all along, and will continue to be.
- As a result, if you’re into that performance and not so keen on brand lift experiments, you’ll feel unloved by the Google sales machine. Don’t be too eager to feel loved, though. Let the performance numbers and the associated profit be all the love you need.
- Test, iterate, and optimize, no matter what your goal.