AdWords
1.8K members online now
For issues or questions regarding logging in, account access, or account security
Guide Me
star_border
Reply

How granular should one go?

Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

Hello all,

my colleague and myself have a difference of opinion and I would like some input from "the crowed" Smiley Happy

His background: data-analyst. Works mainly with databases etc. Mine: Adwords since 2006, relatively little hands on work last couple of years (still enough to support and guide my teammembers that did the hands on work though).

We work in a field that has high yields, but the CPO´s are accordingly. 200,-€ plus.

Because of the nature of the field there are relatively little topkeywords, the main keys consist of 4/6 really big volume keys, and lets say 15 main keys that generate maybe 10% of the volume of the really big keys. BUT as the search-intend is localised we have a huge amount of secundary keys (city names). So that makes for literally millions of keyword combination. (we´re at 3.3 mil. now, spread over 12 accounts) and obviously a massive amount of "low search volume" keywords.

Now to the question: 

I would prefer to give the big cities their own campaigns/adgroups, and bundle all the smaller villages and cities (thousands) in similar adgroups. Thus 5000 keywords consisting of "keyword + city1", "keyword + city2" etc.
As the intend is the same, using keyword insertion I can create relevant keywordtext even though all the keys are in the same adgroup. 1000.000 keys would make up 200 adgroups. If certain kombinations get more clicks, I will take them out of the adgroup and give them their own.

My colleague on the other hand wants to create a ad group for each combination. This means we are looking at over one million ad groups, one for each combination (exact has more keywords in the same ad group).

His reasoning is "the more granular the better". I disagree, and think it makes more sense to have a smaller amount of ad groups. They will be ever so slightly less focused (his one million ad groups would not be using keyword insertion but have each keyword in the ad text), but in my opinion with that amount of ad groups the accounts will just not be manageable.

 

What say ye?

 

 

2 Expert replyverified_user

Re: How granular should one go?

Top Contributor
# 2
Top Contributor
Hi Seamea,

Based on my experience going granular with a 1KW/1AG concept is likely to pay, therefore I'd concur with your analyst. (It doesn't when you go granular in terms of geographic targeting.) The main reason is that you can make the ad copy in each ad group more relevant to the keyword.

Management of such a huge amount of AGs is of course a different issue. Not sure you have the workforce. What we did in particular was an extensive use of our own ad copy creation script via the AdWords API service and that of selecting the appropriate automated bidding options offered by AdWords. We managed to reach 1M registered users within record time and the costs went down significantly.

One more note: be very careful if you choose to use DKI because that's a really double-edged sword.

Well, my 2 cents.

Best,
Lakatos

Re: How granular should one go?

[ Edited ]
Visitor ✭ ✭ ✭
# 3
Visitor ✭ ✭ ✭

Hello Lakatos,

thanks for your quick and comprehensive reply!

Maybe I wasn´t clear enough in my description, but the ad-copy would be almost the same. As title 1 doesn´t have enough room to add more then just the query/keyword (the main keys are long, and so are most city names) title 1 would be just the keyword/query in both scenarios, all other ad text is the same. The only difference would be that for those ads with keyword insertion, for the exact keywords the word order would be the one the user used. As we sell a service, not products, we do not have a huge product catalog where each item would require its own ad copy (in my opinion). The main keywords are just different ways to say the same thing

I realize that in theory it would make the slightest of differences, but in my opinion that advantage would be off-set by the fact that the account becomes almost if not completely unmanageable, thus loosing that minimal advantage.

Example ad text. Most keywords are low traffic. Thus it would take forever to see what ad texts works best. Especially if all those ad texts and impressions/clicks are divided over 1 million ad groups....

 

As far as workforce goes, that would be me. And I´m responsible for everything, so I have to manage Bing (for which we plan to import the adwords campaigns....), SEO and affiliates as well....

Re: How granular should one go?

Top Contributor
# 4
Top Contributor

Hi @Seamea from a performance point of view, there's a strong argument to say that 1 Keyword per Ad Group is the ultimate in relevance.  After all, if there's only one Keyword, you can make your Ad(s) perfectly tuned to that Keyword and no other.

 

However, as @Lakatos said, when you're looking at very large Accounts, you've got to think of practicality as well.  When Accounts are large there are balances that need to made and it's easy to lose sight of large losses when you're finding only small gains.  I've seen several large Accounts where the staff were barely able to do basic day-to-day maintenance and in those cases it's worth considering whether you're getting any real benefit from having such a large number of Keywords.  Now would be a perfect time to reference this recently published Digital Garage article:

 

https://digitalgarage.googleblog.com/2016/08/Is-your-AdWords-account-bloated-with-unnecessary-keywor...

 

So, quite apart from the discussion over structure, I'd personally first be looking at whether you really need all those Keywords in the first place...

 

Jon

AdWords Top Contributor Google+ Profile | Partner Profile | AdWords Audits

Re: How granular should one go?

Top Contributor
# 5
Top Contributor

Hi Guys,

 

In most cases the distribution of performance over a really astronomic number of keywords (or even search terms) follows the rules associated with Gaussian distribution that is also called normal distribution in probability theory. (See: https://en.wikipedia.org/wiki/Normal_distribution)

 

A great many natural phenomena are based on it. That is also what this epic article authored by @Jon_Gritton is based on in terms of mathematical statistics.

 

***   Keywords and the Pareto Principle   ***

https://www.en.advertisercommunity.com/t5/Articles/Keywords-and-the-Pareto-Principle/ba-p/462688

 

Based on the original question posted in this thread, however, it seems that that is not the case with the campaign being discussed. As described, the performance distribution may be a mixture of normal and even distributions.

 

Yea, I am sorry about the workforce issue. I thought more empleyees could be added

 

Best,

Lakatos

 

 

Re: How granular should one go?

[ Edited ]
Visitor ✭ ✭ ✭
# 6
Visitor ✭ ✭ ✭

Thanks to the both of you, for your input as well as for the very interesting links!

@Jon_Gritton: As far as the need for the keywords is concerned, if we would leave out the combination with the city names there aren´t many keywords left. The cities are important/relevant to the product, so including them does make sense unfortunately.

My problem is having them almost all of them in separate AGs, as it leads to over 1 million AG where the only difference is the cityname, which could be covered with keyword insertion. As the product/service we provide doesn´t change, the ad text would always be 100% relevant, and ends up being the same in both systems most of the time.

I´ve read both your articles (your link and the one that was referenced by @Lakatos) with great interest. Coincidentally the number of keywords coincides with the number in our account (3.3 million active Keywords). Conversions are numbered in high 3 digit numbers though, not the 41.000 in your account. By the standards described in your article I should pause about 999.000 Keywords/AGs, which brings me to my 1000 AG as well Smiley Happy

Re: How granular should one go?

Top Contributor
# 7
Top Contributor
Your situation is pretty complex to summarize anything, however, I think the best option for you would be to promote as many keywords to their own AG as you can comfortably manage, with or without scripts. As I mentioned earlier DKI is a double-edged sword for which reason I'd prefer discouraging you unless you have enough experience to cope with advanced issues. Else you may want to give it a try with very limited scope.