There are some really great things about Google’s ad display algorithm and there have been some VERY significant changes made recently. Typically their changes with which keyword/ad combination is displayed are good… this time, not so much.
So, I’ve already discussed the Five Mill Tree Method, but it was a high level explanation. In order to do the actual “slicing and dicing” of keyword traffic, you need to have the tools to make this possible. Unfortunately, Yahoo doesn’t give you the tools to effectively and efficiently do so, but Google does.
Let’s first go through a couple quick basic rules of Google keyword matching…. and then I will explain how Google just recently made it better, then made it worse.
Match Type “Trumping”
Let’s say you are running:
- 2 Identical Ads
- 2 Different Match Types
- 2 Different Bids
We are dealing with 2 different Ad/Keyword/Match combinations.
#1
Keyword: socks
Match Type: Broad
CPC: $3
#2
Keyword: [socks]
Match Type: Exact
CPC: $1
Let’s now say that a user searches the query: [socks] (Exact match)
One might assume: The match type for combination #2 is the same, but the bid for #1 is the higher. Due to the fact that both match types could address the query [socks], the corresponding bid/combination would be the higher bid; the higher bid broad match has cannibalized the traffic. This assumption is entirely WRONG. Google associates the keyword of closest match with the query. In other words, first Google notes that there are multiple keywords that are the same (but different match types) and matches the one with the most restrictive match type. In this case, the keyword is [socks], and the bid corresponds to the lower bid because the match type is most restrictive.
Geo-Target “Trumping”
This time:
- 2 Identical Ads
- 2 Identical Match Types
- 2 Different Geo-targets
- 2 Different Bids
#1
Keyword: [socks]
Match Type: Exact
Geo-Target: USA
CPC: $3
#2
Keyword: [socks]
Match Type: Exact
Geo-Target: California
CPC: $1
User Query: [socks]
User Location: San Francisco, California
Similarly to Match Type “Trumping”, the query is mapped to combination #2 because the Geo-target is most restrictive, despite the lower bid. No canibalization occurs (but it would cannibalize with Yahoo!).
There are a few caveats, but, for the most part, the above is correct. Two major changes have been made, one for the better and one that is not.
Change #1: A Keyword Below Minimum Bid Threshold, is Still a Keyword
Previously, if your keyword was below the minimum bid threshold, it would not compete in the bid auction. For example, let’s say that in combination #2 in the above examples, the bid was below the minimum bid requirement. If this were the case, Google pretends that the keyword doesn’t exist and it would map to the higher bid despite the keyword of closer match. This change NOW takes into account keywords that are below the minimum bid threshold.
This change is fantastic. This gives us the opportunity to target keywords in as much granularity as we desire and bid based on our metrics of success. We no longer have to worry about the effect of a keyword dropping below the bid threshold.
Change #2: Google Matches Keywords When it is Better “Quality”
So, everything I’ve said above blows up (although very marginally) based on this new caveat that Google adds to the equation. The deal is that everything above is true, EXCEPT in the case when a keyword that is not a “closer match” has a lower max CPC and higher Bid x Quality Score (essentially Bid x CTR) value .
This is how Google says it:
”
On rare occasions, the system will prefer to use a keyword that is cheaper (i.e., it has a lower bid), has a higher Quality Score, and has a higher Ad Rank. Here’s an example:
Query: plumber tool
Keyword 1: plumber tools (maximum CPC bid = $0.10, Quality Score = 30, Ad Rank = 3.0)
Keyword 2: plumber tool (maximum CPC bid = $0.15, Quality Score = 4, Ad Rank = 0.6)Ordinarily, Keyword 2 would be preferred because it matches the query more closely than Keyword 1. However, Keyword 1 is cheaper, has a higher a Quality Score, and has a higher Ad Rank. Therefore, the system will prefer showing Keyword 1 in this instance.
Keep in mind that Quality Score is calculated every time your keyword matches a search query — that is, every time your keyword has the potential to trigger an ad.
”
As a result the ability to be extra granular gets thrown off, because at anytime Google can match your less granular keywords with queries when it gives Google an opportunity to make more money. (You can word this sentence however you want, but it will mean the same thing.) Google will argue that you are getting a chance to appear on a better performing keyword combination.
So, here is where Google entirely fails.
1. When you take away an advertisers ability to be as granular as they want to be and when you make advertising less clear, the ability to advertise most effectively gets impaired. The venn diagrams that you see in my previous posts are now better drawn with dashed lines because it is now impossible to be definitively granular.
2. The person/people at Google that came up with this one may or may not have thought they were doing advertisers a favor, but the reality is that they have created extra revenue for Google in the short-term and that’s all. Personally, I advertise with so much granularity and analyze keywords at the keyword level, that changes like this will only make my advertising less effective. In fact, any advertiser that is scientifically managing their bids will be hurt by this change. By making advertisers less effective, advertisers cannot optimize for metrics of success as scientifically as they were previously able to do. Assuming that advertisers are advertising with financial efficiency (which is a gross assumption with many advertisers), the long-term effect, by definition, is a loss of revenue for Google, a devaluation of Google advertising, and a financially negative effect on its advertisers (until they realize that less money should be spent on Google).
Keep in mind, though, that this change is very negative, but it’s effect will be barely noticeable. But, in “advertising equilibrium”, as I like to refer to it, it is bad for everyone.
In short, I challenge anyone to find an example where this change actually benefits advertisers…
Ps. Check out this link: http://adwords.google.com/support/aw/bin/answer.py?hl=en&answer=66292 to read more information about Google Keyword Matching
Pps. in light of this post, i’m adding a parent category and a sub-category: “Google”, “Fail”
] coined “The Five Mill Tree Method.”
Follow Me!