Traffick - The Business of Search Engines & Web Portals
Blog Categories (aka Tags) Archive of Traffick Articles Our Internet Marketing Consulting Services Contact the Traffickers Traffick RSS Feed

PPC: Fast and Slow

Posted July 11th, 2013 by Andrew Goodman

At our house, “Canada’s Worst Driver” is a huge hit. The show is so popular nationwide that it’s into its eighth season. Not too shabby.

Speaking of “hit,” that’s what contestants do to the inanimate objects in the obstacle courses, on a regular basis. On “Canada’s Worst Driver,” the expert judges consist of a retired cop, a couple of advanced driving instructors, and – yes – a psychologist.

This last one is telling, because the show is like a lab demonstration of just how sound theories of attention are. Even if you’re a talented driver, distractions can be hazardous. Distractions can drain anyone’s ability to concentrate. The most dysfunctional drivers also happen, quite often, to be part of dysfunctional couples. When the hyper-critical spouse is riding shotgun, there is screaming, crying, and denting of fenders. When they kick the spouse out and let the “hopeless” driver go it alone, there is often marked improvement.

Many of us would rather rely on our “raw skill” than admit we need to step back and build in a consistent routine intended to manage our natural limitations.

In the highly acclaimed “Thinking: Fast and Slow,” Daniel Kahneman synthesizes decades of research that has uncovered two distinct thinking “systems” in our brains. System 1 (reactive) is essential for bundling together eons of evolution and a lifetime of experience into seamless performance. As amazing as this system is, its performance is easily degraded – for example, by fatigue and distractions. It won’t work effectively in the modern world without System 2, a deliberative form of thinking that is required to produce rational outcomes.

Back to the example of driving: assuming we’re reasonably skilled drivers, Kahneman notes that we’re amazing at letting our eyes and hands work to steer around a curve, with little consciousness of our skill in processing the information and reacting to it. But we need System 2 to tell us to set up a routine like drinking coffee, singing, and rolling the windows down on a long drive. Although unromantic, years of deliberations by engineers have taken our “skill” of slamming on the brakes in dangerous situations and created a much safer (machine-driven) stopping method: ABS. Neither our legs nor our brains can jam the brakes on and off fifteen times per second. We’re safer now on slick pavement because of System 2; i.e., what the engineers built and tested over a long period of time.

Too often, we hop into our PPC accounts with an excess dependence on an all-too-human System 1, like we’re some spreadsheet-enabled Tarzan swinging on a vine.

Neither being busy, nor being “really good at this stuff,” should be an excuse for completely ignoring the need to incorporate deliberative, “slow-thought” protocols into your campaign management methodology. The fact that it feels “right” to nervously tweak incremental details of accounts during the “fast season” doesn’t make that activity particularly effective. It doesn’t mean you’re off the hook in terms of System 2.

Now that we’re hitting the summer months, you may be relieved to have a little breathing room to work on more methodical, “important but not urgent” initiatives. Indeed, if you put in the time now, things might feel less frantic come fall. Not only that, but System 2 is great at building systems, protocols, and machines… not just operating them with a twitchy trigger finger. Systems, protocols, and machines are what will make you the real money. How are you going to build something new or better that scales, that relies less on your raw animal (or even advanced cognitive) abilities?

Here are some to-do’s to consider:

  • Set up new remarketing audiences and come up with new image ad creative. If you have a PPC account with little or no remarketing set up, you need to do this yesterday. The same goes for replacing outmoded remarketing code with “new” remarketing code, such as Google’s Universal Remarketing Tag. Remember that a remarketing audience must be cookied as such; your audience size starts at zero. Get that code installed on the website! Start now! If you’ve already started remarketing, work on thinking it throug more thoroughly, with custom combinations that include cart abandoners, a “don’t show any ads for the first three days” setting, etc., as seems appropriate to your custom needs. Or contact your Google rep for more information on Dynamic Remarketing, if you retail a large number of products. Finally, think through what story you want to tell, or what brand image you want to portray, with your image ads. You’ll be serving tens of millions of impressions, typically. What do you want people to see (over, and over, and over)?
  • Build simple or complex bid rules and figure out how often you’ll run them. Along with the more obvious account-wide sweeps you might do to look for anomalies or opportunities, you can also try less comprehensive, non-obvious ways of building filters and bid rules to create bespoke outcomes. For example, filter for match type and bid down on a match type you feel has been grabbing too much impression share away from better match types. Or run an account-wide match type performance report using a third-party tool like Optmyzr. That can be a nice way to inject your creativity and your “rock star powers” into the equation. But it is System 2 thinking because you’re slowing down and taking stock of how the account functions, instead of merely reacting to little pieces of data like Miguel Cabrera diving out of the way of a brushback pitch.

brushback pitch

  • Review long-running ad tests with insight into why they were set up that way in the first place. Don’t just stab the pause button at lower performers, and don’t end tests that are too close together in performance yet to be statistically distinct. Take some time to recall what principles went into the tests, and if you draw any conclusions, take time to communicate those with someone – preferably in writing. To take an example, a client in the office furniture business has several internal landing pages that are underperforming the home page in terms of ROI on PPC. Had we followed only best practices, we’d have stuck rigidly to the notion that the keyword should take us to the focused landing page, and that is that. We now have more insight into the workings of the site as it relates to different product lines. Testing is about the spirit of inquiry, which requires continuity and planning. It’s not nearly as effective when it’s only about stabbing a pause button.
  • Deliberately incorporate teamwork; don’t be a lone wolf. Successful testing, and successful companies, require a culture of acceptable debate and disagreement, as shown by Jim Collins in “Good to Great” – Collins calls it “confronting the brutal facts.” Testing isn’t about consensus, it’s about learning and iterating. As Seth Godin pointed out in “Survival Is Not Enough,” you need to think about how to involve various perspectives in your decision-making so that there is a robust mDNA (meme DNA) in your operations. Godin insightfully argues that in terms of evolving to meet new industry challenges, “competence” can be a company’s worst enemy, as it gets stuck on a routinized “winning strategy.” To shake things up when seeking input, consider a shortcut: informally “crowdsourcing” within your own company (assuming you’ve encouraged diversity in your company) – use an anonymous process if that helps. Be supportive of “weird” ideas, as long as they’re not the only ideas people suggest.
  • Get high-powered tools working for you. Above, say, $500,000 in annual ecommerce revenues, do you really have any excuse not to put a program of A/B testing in place for key landing pages? And for a busy home page driving $2 million or more, is there any excuse not to seek out an advanced multivariate testing tool to see if you can’t create a lift of 15-20% in conversions from visitors landing there? Simple math: 15% X $2 million = $300,000/yr. in increased revenues. For that kind of money, it’s pretty obvious what you should do: push away from the desk so you can’t tweak those same keyword bids 50 more times, or add another 700 negative keywords to a campaign. Get the right tools and team together and set that goal to generate that 15% lift in conversion from that page.

Cognitive scientists like Kahneman and his colleagues often employ a simple word to describe our tendency to over-rely on System 1 (reactive/heuristic thought), and to avoid too much engagement with the more mentally taxing System 2: “lazy.” It seems we’re hard-wired to conserve mental energy. Reacting does not wear us out as much as stepping back and planning. But of course a day spent reacting will be a day where we eventually make many mistakes as we fatigue, even if we’re “smart” or “sharp.”

So, plan we must. And some of that planning needs to create the kind of consistency that appears to take “us” out of the equation. By taking steps to remove your System 1 self out of the equation more often, and using your System 2 self to do so, arguably, it’s a higher-order “you” that’s involved. “Smart” or “sharp” needs to give way to “rational” when dealing with large, complex systems. It is rocket science.

To profit in the midst of modern, complex systems, our natural aversion to System 2 thinking (it’s just so taxing) is something we need to combat. System 1 was great for prehistoric man running from a hungry predator, and remains an advantage in the sport of dodgeball. But it’s terrible for sitting at a desk and correctly intuiting statistical confidence in an ad test.

Schedule an appropriate amount of time in your schedule for System 2 thinking. There’s no time like the present!

An earlier version of this column ran at ClickZ on Nov. 30, 2012. Reprinted by permission.



Comments are closed.


 


Traffick - The Business of Search Engines & Web Portals

 


Home | Categories | Archive | About Us | Internet Marketing Consulting | Contact Us
© 1999 - 2013, Traffick.com. All Rights Reserved