Confession of a Googletizer

Let's face it. Google dominates the pay-per-click market and is THE benchmark for search engine.

Translation: if you engage in any manner of online marketing, you are a Googletizer.

Like me.

I used to have very warm and fuzzy feelings toward Google but I do wonder if the recent success, financial and otherwise, of Google is turning it into your typical hi-tech bumbling hydra since there are so many things going on in those campuses.

My last post noted Google's attempt at dictating how a word/expression should be used (whitepaper, bad; white paper, good). Now, I noticed that my ads are showing up at unexpected places. As a Googletizer, one of the main value of Google service vis-a-vis other PPC vendors is that Google seems to be pretty good at keeping its words and leveling the playing field. So, I was very surprised when my ad shows up at a location that looks like a link-farm - a website that offers nothing but links where the site operator gets paid when a viewer clicks on a link. However, as a Googletizer, I am most adverse to link farms because it does not provide high quality clicks and the operator has incentive to create click-fraud to boost their own earning. (The Economist even noted operations in China where people are hired to click on these links to generate money.)

For this and other reasons, I have opted opted out the Content Network option with Google to minimize this kind of exposure. So, imagine my annoyance when I found a click that come from a site that is a link farm (no search capability and no meaningful content). I filed a request to Google support asking for clarification of policy and how to avoid this kind of exposure. I got a reply assuring me that Google does its utmost to have the right content site operating for my ad's. I then replied with "but I have opted out of Content network." Google gets back with a comparison between the Content and Search network and advise me that if I wish to opt out of a specific site, I can do so from the Content network option. (But, I have opted out Content network completely?!)

Anyway, suffices to say that I am not impressed with this kind of service and find their black-box process wanting. I am perfectly willing to accept the argument that I have stumbled onto a special micro-site run by Google to improve my performance. that would have been kind of cool... But, at least be forthcoming about it!

==
Anyway, I recently read David Ogilvy's Confession of an Ad Man. He is my new hero. He is nothing like the average advertising types that I know - thank god!

Anyway, my latest experiment with Google is to separate out the winners from the pack and give them their own campaign. The idea being that these VIP's will get special treatment with better budget allocation to ensure good ad placement. For the rest of the pack, I will just use the auto-budget option.

The auto-budget option is not a bad thing, if you do NOT accept any of the Google suggested parameters. There is nothing inherently wrong with Google's suggestion. But, Google gives you the average and, unless you only aspire to be average, you should not take the face value.

Yes, my grand theory of inter-temporal black box optimization. I had some additional thoughts about it over the past few days when things are less crazy. I think it comes down to the ability to set out a systematic program to optimize from component level to system level. For example, selecting the keywords is the typical first step. Then, use the web log to optimize the keyword selection. Then, there are all sorts of tricks for keywords in terms of capitalization, order, and matching options. Then, the typical second step is to optimize the bid price. I have found it usually take a few days to get a good baseline and it is a constant adjustment between auto-budget and manual control.

I think one of the key points in the process is figuring out how to know when enough is enough. For example, how do you decide that a keyword/ad is a dud. For the longest time, I was thinking in units of time. It is a good proxy for similar campaigns. However, for a brand new campaigns where there is no basis for comparison, time is a tricky thing. Somebody suggested that I consider the number of impressions. This is actually quite an enlightening idea for me, I know - it does not take much to impress me. So, instead of waiting for two weeks, it has to wait for X impressions. Rumor has it that 1,000 is the magic number for Google engine. Of course, one early caveat is that if your keyword selection is rather poor, it may take forever to get to the 1,000 mark and, consequently, risking introducing too much temporal distortion into the analysis.

Speaking of temporal distortion, how does viewer/user preference change over time is another topic that I find fascinating. The key question being that how do I go about capturing the change at the earliest possible time? For example, in the old days (of course Al Gore has not invented the internet then...) people may search for a "personal computer" but at some point people would search for "PC" only and these days people probably only search for "Dell". So, how do I capture that shift in viewer/user behavior?

Another experiment that I am embarking on is SEO, search engine optimization. The idea being that PPC is still relatively expensive. And, as viewers sophistication increases, the legitimacy of a high natural search ranking is significantly higher than the paid-for high ad ranking. I have engaged in some SEO on the corporate site before, but it was a bit clumsy and difficult. My latest insight is to break the process into compartmentalized pages. So, instead of optimizing for the top five terms that I want for index.html, I can have five individual pages optimized for these terms and have it linked into index.html. On the face of it, the idea makes a great deal of sense and I am hoping to get these SEO pages implemented in the next week or so. This would indeed be a very exciting experiment for me.

==
Happy 2006!

chiefchickenheadless (at) gmail (dot) com sign out
Legend: (at) = "@" (dot) = "."