Looking at the right stuff but looking at it from too far away might be hindering your optimization capabilities.Â At this level of optimization, the factors that you need to consider are much more complex than campaign level CTR and CPC.
For those struggling to find a meaningful test environment, I would suggest doing the following before moving forward with a SEM test.Â This will help you gather data and possible outcomes.
First of all, for a test to be feasible you need two things:Â high volume traffic and client (searcher) interest.
In order to find which KWs would be suitable for a test you should look at the campaign at the pure KW level. Forget about campaigns and adgroups, those will come later.
Here are some guidelines to help you build a [better] test:
Don’t bite more than you can eat.
By that I mean, take a smaller part of the account and optimize that, rather than optimizing the entire campaign. That way you will be able to prove that the optimization and testing is working by comparing past performance to your current test.
High performance is paramount.
Analyze high performing keywords at the keyword level not at the adgroup or campaign level. Group them together semantically.
At this point we have a very good idea of the topics (aka adgroups) that we are running and we know the hierarchy that we should be following in order to support the high level business goals.Â So don’t worry about the adgroup/campaign naming conventions or structure. That will come later.
Use the top of the pack.
Take top 20% of high volume and high conversion keywords and do some competitive research to see what is the possible volume of traffic the testÂ might get.
Price is a key indicator.
Look at the CPC numbers and search for the high cost terms. If the keyword is not an industry standard such as CRM and the volume/competition is not that strong then you know right there that we might be overpaying for that KW.
Group terms smartly.
Take all the keywords that fit that criteria and separate them into adgroups no larger than 7to 10 keywords. The grouping should be done at the keyword level semantically not at the topic level. What I mean is that if the keyword has the word database, then find seven terms that have that same term on it and group them together.
Assume that you are ALWAYS overpaying for Google clicks.
Find the bidding sweet spot.
A balance between too many impressions and clicks and not enough.
Use your tail.
Take long tail keywords that have strong conversions and some clicks and add those to the campaign.
This is your chance to expand the keyword list. Do not go crazy. Add a max of 20% new keywords every few weeks. So if you have 100 total you have at most 20 new terms that you can add. Be smart about them don’t just add anything the keyword generation tools offer you.
Superstars and super-spenders.
Big difference. If you find a superstar term (i.e. CRM) or a high volume keyword that you are not sure about (i.e. high performance) separate them from the rest of the campaign and put the into their own adgroups.
After you do all of these (and your own stuff) you can have a good idea of what the optimization test would look like.