|Google Stability Index Number, Adwords style|
Checking Adword dependency
| 6:54 pm on Aug 2, 2006 (gmt 0)|
Now you will have to bear with me. Itís an idea Iíve got, and Iím still not convinced it will be possible, maybe with a little help.
In this post [webmasterworld.com...] ronburk talks about something he calls The Google stability index number that at first glance looks complicated, but actually is not that hard to work out. The index number says something about how dependent/vulnerable ďAdsense professionalsĒ are towards Google fluctuations.
What Iím hoping fore is to make the same with Adwords. A google stability indesnumber, that says something about how dependent/vulnerable ďAdword professionalsĒ are towards Google fluctuations.
And maybe taking it further and also look at YPN, MSN ads and other ppc pograms.
Anybody into the math and able to spell out the formula.
| 11:16 pm on Aug 2, 2006 (gmt 0)|
I guess the simplest measure would be to just sort your AdWords-supplied traffic by ad. You would then basically be checking to see the smallest percentage of your ads required to supply half your AdWords-supplied traffic. Smaller meaning worse, of course, since if 1% of your ads is supplying half your traffic, than a Google price change (or a change in customer search habits, or the loss of a high-volume Content publisher your ads are appearing with) only has to wipe out 1% of your ads to wipe out 50% of your AdWords-supplied traffic.
Seems like the next step up in complexity would be to factor in the price per click you're paying for each ad, along with the ROI of that ad. The idea being that if you're dependent on ad X for 20% of your traffic, that's not quite so bad if you can still turn a tidy profit even if you have to pay 3 times as much per click. I.e., just how much does Google have to raise the price on an ad to wipe out its benefit for you?
There's also some added complexity in that ads probably aren't really 100% independent variables. If I'm bidding on "antique widget" and "1938 widget", is the risk of lost traffic or increased price for one of them really completely independent of the same thing happening to the other?
But, even the very simplest measure might be useful just to inspire you to expand your ad campaign so it's less dependent on a small minority of ads for its success. And to see if you're getting more vulnerable or less vulnerable to Google changes over time. Just seeing that Zipf distribution of results for the first time can really alter how you allocate your efforts.
| 12:03 am on Aug 3, 2006 (gmt 0)|
I would have imagined that advertisers would be safe from the uncertainties of algorithms until this late unpleasantness set in. The method you linked to in your post seems like a good defense against a random loss, but from the reports of advertisers who were shut out of their campaigns seem to indicate that there is some (admittedly opaque) criteria for site evaluation. It's like there's a checkpoint and the guard asks you "Who is Mickey Mouse?" before he lets you through; it doesn't matter how diverse your campaign is because it isn't a random event.
Anyway, a lot of MBA folks seem to like to throw around the idea of the 80-20 rule, which states that 80% of your revenue/traffic/sales/enjoyment will come from 20% of your customers/advertising/sales force/activities. They apply this rule to all sorts of things professionally and personally in order to make themselves more efficient. If you find that five of your keywords are generating the majority of your revenue make sure that you are getting the most out of them. You should diversify, of course, because the market may change and you need to know that and already be there with your next five (or ten, or a thousand); but first make sure that you're living in today while you're preparing for tomorrow.
Finding your top keywords shouldn't be too hard, just get a list of your keywords and sort them by revenue (or whatever you want) start at the top. If you really have a very even distribution of revenue over your keywords, then you may not have enough keywords in your set.
[edited by: MikaelTC at 12:10 am (utc) on Aug. 3, 2006]
| 1:26 am on Aug 3, 2006 (gmt 0)|
|I would have imagined that advertisers would be safe from the uncertainties of algorithms |
I know a lot of advertisers eschew the content network, but some of them must get significant traffic there, 'cause a whole lot of money is being spent on clicks. So it seems likely that some advertisers are reliant on some small number (Zipf distribution kicks in again) of content publishers for their traffic, and it follows from that that they can be affected by the regular ranking algorithm changes, not just changes on the AdWords side of the house.
For such folk, it might be more interesting to analyze how stable of a base of publishers they have that can fulfill their ad inventory.
| 7:26 am on Aug 3, 2006 (gmt 0)|
Hi Ron and Mikael
Thank you for your ideas and insight. My own ďalgorithm developmentĒ followed the outline of Ronís first post, but stopped just short of adding the price per click factor.
How would you deal with smart pricing (a subject Iím not sure I understand completely just yet) would you need to take this into considerations or would you just look at the price as a number regardless og smart pricing or not?