Is there any reason not to do this? The only one I can think of is if Google will make broad match decisions based on the entire pool of keywords in the ad group. That could be beneficial. Does it work that way?
Does it work that way?
Not so far as I am aware - and I am reasonably aware. :)
Now, of course, I worry that eWhisper is going to post and say I am mistaken, based on his several years of exhaustive testing, heheh.
It scares me that he knows more about AdWords than I do. :)
I will leave it to others to comment on the goodness (or not) of one ad group and one ad per keyword in general terms, based on their experience.
This quick note, however: if you happen to be targeting the content network I think that a really tight group of up to 20 keywords which are all highly focused a single topic would be much better (in terms of delivering your ads to pages about that same topic) than just one keyword.
My $0.02.
AWA
i agree with AWA on content, but here's a search perspective.
optimization requires imps and clicks. bu dissecting your machine into such small pieces, you're removing the advantage of aggregated behavior (for like keywords) so you'll be slower at ad optimization.
long tail has an issue, G won't show ads for keywords with low amounts of searches, so you're going to have many ad groups that get stuck in that no action pit. sure, if they were commingled within an ad group they'd still get idled, but there's this trickle thing within adwords, where they need to occasionally let things through so their algorithm has data to make decisions - i think by shredding things to bits, you're missing some trickle that could help you, both for low imp keywords and certainly for the ads that would be better tested with more data on each.
also, i see a tendency in technical minded people to read too much into datasets, the problem here being that finely dissected data seems more meaningful to many tech minded folks... i think my making all of your observations (that are required to make sound optimization decisions) microscopic, you'll be focused on the trees and miss the forrest. this is where clever techies point out that they can reassemle the granular data so it is meaningful in the aggregate, but like i said above, it isn't aggregate behavior, it aggregated microscopic data pieced together to appear to be a larger dataset. in my experience, this leads to analysis problems that causes more frequent misdiagnoses of the mechanisms of what is happening. an analogy... a rain gauge is great for measuring rainfall, it does have shortcomings that must be considered. however, compared to teeny thimbles being spread out, then aggregated mathmatically, it is far superior. consumer behavior viewed too granularly divert focus from what they are thinking and doing, and the same is true for how G reacts to your generated data. somewhere, there are genius geeks within G writing algorithms, and the write code to analyze granules (ads, keywords, pages), but they also write code to try to comprehend non-granules / chunks (ad groups, campaigns, websites)... if you wholly depart from chunking behavior, i believe there will be a mismatch between their algorithms and your efforts that will result in suboptimal performance for you. it's better that dissect data for analysis than it is to aggregate. aggregating it assumes the behaviors are the same across granules, that's almost always an incorrect assumption. dissecting aggregate numbers has shortcomings too for sure, but you can test those assumptions by making modifications... if your approach is microscopic, there far less changes that can be made.
esoteric discussion for sure. as an analytical person, my best suggestion is for you to bracket some data and try both approaches and see which works better for your circumstances. don't choose to be granular or chunky or super chunky or ultra clumped, test variations of them.
ultra clumped is how i often find new clients... one campaign, one ad group, one ad... and i give them advice to move towards microscopic - the opposite of what i'm telling you. so i'm a bipolar liar at times too, but i try to be honest about it, i think. :-)
shotgun or rifle marketing approaches are usually the choices i see being flexed. shotgun is too broad, you get slaughtered for irrelevance in adwords if you try to be a shotgun. rifle is far better. but i think one keyword - one ad - on ad group goes too far beyond rifle, closer to a tunneling electron microscope.
[edited by: RhinoFish at 10:45 pm (utc) on May 8, 2009]
