If you headed over to Brighton SEO this September, you’ll have seen that the keynote speaker was Gary Illyes at Google. Jennifer Slegg asked some really good questions during the keynote Q&A! Though as Gary can’t really give away much, there were not very many actionable things you could take away. Other than it is pointless to ‘chase the algorithm’, off the back of the comment that 95% of algorithm updates are not actionable.
I walked up to Gary after for a chat, and managed to get a picture too, but also attempted to ask some questions and listen to others questions. The answers he gave were a little more detailed than what was stated up on stage, so it was a real good bit of insight that most others didn’t get. Gary was previously an engineer for the search ranking team, so he does know a thing or two, but whether he gives anything away is another detail however.
Does AdWords interaction have any sort of impact on Organic Search?
My question to Gary was around AdWords impacting Organic search. At first when I asked this question, he stopped me before I could complete the question. The question was not whether the AdWords team interacts with the Search engineers, but rather, do the Search team at Google recognise a relationship between AdWords behavior impacting/influencing click-throughs to organic results.
Gary’s answer was an immediate no to AdWords impacting organic search. I had to clarify and follow up with a more detailed question off the back of that question, which I did..
There are not too many public cases where site owners have given away the fact that AdWord results has had an impact on the results in organic search, with specific regard to the brand in question appearing. There is a case study with eBay, in fact, from 2013 showing the CTRs for brand searches, and what the impact on organic CTR looks like when turning off AdWords. You can guess that it increases, as people will look for the brand in the organic results. But what about for generic results, and how does impact rankings over-time? The interesting thing for me is that Gary mentioned at the conference that Rank Brain, the mysterious machine learning algorithm, is an algorithm that looks at “historic trends” in Gary Illyes’ own words.
As an aside to this, outside of the Q&A on stage, I was with others asking questions, and whilst I did not bring the up the question, he did bring up the terms ‘head tail’ and ‘long tail’, which makes me think that Google categorises their data sets in this way and they treat these two categorisations separately. So this got me thinking, in an a/b test, in order to get statistical significance, you need enough data in order to be happy with those test results to apply them to 100% of traffic. That statistical significance gives you comfort in knowing that a good proportion of people liked a change that you made. And whilst Google says they do not use CTRs to directly rank websites as the metric is too noisy, they use it in AdWords to conduct tests as a measure of success and SEOs tend to use it to conduct their own tests, so why would they not use it? Manipulation is one answer to that, but AdWords voids manipulative clicks, so I am certain an algorithm of that nature could be applied elsewhere in search.
This leads me on to the part where I think an area where Google can be using CTR in conjunction with Google’s RankBrain machine learning algorithm. If it can look at a trend of say, clicks going down on a result for a head-term that has statistically significant amounts of searches, then it can deduce over-time that the result has not been as satisfactory for a percentage of what it has categorised as its ‘trusted’ user base, a subset of users that it has determined to be a statistically appealing for its audience. We know Google are a smart bunch, and they do lots of things to tweak the algorithm, and it would be silly for them not be using data in a way that could inform their search engine like this.
And lastly, this leads me to the point that if brands are spending more in AdWords, then they’re harming the click through rates of their organic results, as they are lowering what would actually be higher click through rates.
Gary admitted that the core Search ranking team at Google and the AdWords team do not interact in any way with each other and even said that he’d had to escalate cases where someone in the AdWords team had tried to get in contact with him.
Effectively, this means that if an algorithm or factor like RankBrain is looking at this data to inform ranking, then overtime, if your CTR is going down, this is looking more and more like your website or brand is less appealing to Google’s users. In which case, if it goes lower than other brands, who statistically would pick up more traffic in those positions, then why would they not rank other sites over that site that is seeing the decreased CTRs.
Whilst I’d say that the impact on rankings overtime with regards to a lowered CTR, is purely a theory, the lowering of CTR due to AdWords spend is fact. To end with, whilst this has just been a big block of text of mainly my thoughts, I think we need to take what Google says with a pinch of salt, as of course, they are not going to give away much. It’s undoubtable that AdWords has a huge influence for big brands for generic terms, and with the lack of communication between AdWords and the Search ranking team, I fear there is a huge element that is purposefully being ignored for the benefit of Google’s bottom-line. I’ll end on a somewhat high by saying that whilst not much can be done, continued testing on this would be the appropriate step, but I think as brands get smarter, more and more in-house staff particularly will notice this as Google tries to enhance its AdWords proposition.
I’ll end with the picture I mentioned at the beginning of this post with Gary Illyes:
