<< by on June 11th, 2013
On this hazy and shockingly chilly morning in downtown Seattle (my body is still adjusting since it’s not in the thick Charleston air)I arrive a bit late to the first session due to a crazy-long registration line, but my day gets off to a running start with a really technical session about analyzing PPC data. I love it!
Learn some tips from highly reputable men in the industry, from both Adobe and RKG. Enjoy!
Chris Haleua, Product Manager, Adobe Media Optimizer, Adobe (@chrishaleua)
Chris starts off the conference by discussing the best ways to look at your PPC data to move mountains quickly. First things first, be sure to take metrics with zero data out of the mix. They are going to water down your data and if you don’t filter the zeros out then you are basing decisions on data that doesn’t really matter or maybe cutting keywords too short before they have a chance to gather data. Once you filter out the zeros, be sure you sort by the next best metric. If your end goal is to reduce CPA or ROAS, you not only need to make adjustments based on the CPA/ROAS metrics you’re seeing but also look at other metrics that may have an impact on that final metric. If you focus on making small tweaks pretty often on a large scale, you’ll have more success in moving the metrics that matter in the right direction.
Kohki Yamaguchi, Sr. Business Analyst, Adobe (@kohkiyamaguchi)
While joking about only search marketers being crazy enough to gather early in the morning just to listen to mad scientists Kohki jumps into Quantifying Quality Score.
The Ceiling Effect occurs when a keyword should realistically have a quality score of over 10, but per the formatting of the quality score scale, can only be capped at 10. This skews the overall quality score a bit, and there’s not much we can do about it, but it’s something good to be aware of. While it’s a good model to score relevancy, it’s not perfect.
Through much analysis, the relationship between CTR and Quality Score can be pinpointed to an extent. After analysis, in general a 20% change in average quality score requires a corresponding 27% change in CTR. This is a highly generalized concept and if you can take into consideration external factors and do this at the keyword level, the more probable it will be the case in your situation. While there is a very convincing correlation between CTR and Quality Score, that correlation will differ for each advertiser.
So how does Quality Score affect CPC? If you do an adequate amount of analysis, there are ways to figure out exactly what needs to change and by how much to have a positive impact your Quality Score. After much analysis and various models, Kohki has concluded that there is a 5-10% reduction in CPC for every unit increased in Quality Score.
All-in-all, after you figure out the trade-off models for your efforts, then you need to sit down and plan out how you can do that. Sometimes the solution will be simply doing some ad testing, landing page testing or bid adjustments.
George Michie, Co-Founder & Chief Marketing Scientist, RKG (@rimmkaufman)
George Michie of RKG discusses Product Listing Ads (PLAs) & Cannibalization. So, to address the question of is there some degree of cannibalization due to PLAs? In a word, “yes.” It is going to take some of the traffic that would have gone elsewhere (like organic or other paid results). While some cannibalization exists, people are more likely to click on those listings (CTR is typically higher for PLAs than other paid ads) since it’s highly targeted and displaying exactly what they are looking to buy. George has looked into the theory that ads that show in positions 4+ have seen the most cannibalization since those ads are now less prominent when PLAs are displayed. Out of all the data he looked at (CPC, CTR and actual traffic from paid search), the percentage of text ad traffic drops when PLAs are present for ads in positions 4+. This is quite logical if you think about it; there’s a big block of ads in the space where ads 4-6/7 used to show, and now they are pushed down lower on the page, a lot of times below the fold and there are even some ads getting pushed off the page altogether.
With his data set, while more traffic is coming from PLAs, the CTR from all ads total is actually growing when you include PLA data into the mix. Playing devil’s advocate, are you just getting ahead in one area because you’re taking away from another area, is that really an improvement? George suggests looking at this like a venn diagram; yes, there is some overlap, but yes there is still also some incremental traffic in spite of a little cannibalization. This is the case when looking at text ads vs. PLAs, and also organic vs. paid search. Many clients worry about this when launching paid search efforts, but get them to think of it in this way, and better yet, gather the data and show them while they are indeed losing a little bit of organic traffic or text ad traffic, in the end, it results in incremental traffic.