Quote:
|
Originally Posted by diditcom
Thanks Danny for the post and link to one of my columns on portfolios. I'm not as frequent a participant in the forums as I should be and will try to jump in more frequently. Some folks like to sell portfolios or a portfolio-style campaign management technique as mystical or magical. It's math. All decisions made about bids (individually and in groups or portfolios) are based on some kind of formula. Yahoo's Panama will even bid manage if you want it to.
In the end, results matter, some formulas will work well; some will be killer (especially when combined with a smart team and sound strategies). Others will fail to reach the highest efficiency. Just like in the stock market historical data does not always predict future behavior or results. Some formulas don't adequately take the unpredictability and volatility of certain keywords into account. There can be smarter portfolios and simple (dumber) ones. The addition of non-bid parameters into the equation (CTR and fuzzy quality scores) make it more challenging for the developers of formulas to build the best formula or AI.
So, buzz-words aside (such as portfolios, smart portfolios, predictive models, etc.) are really just marketing speak that attempts to name a formula, a methodology or a philosophy.
I've found that one formula doesn't fit all. A flexible AI that can test different formulas and a team of analysts that can adjust formulas and campaigns will often outperform a "one-size-fits-all" formula (not that anyone out there is using such an inflexible formula).
As the engines continue to change the way they calculate predicted CTR, predicted Ad Relevance and predicted eCPM (yield), those of us tuning and tweaking formulas will have to keep on our toes. None of us will have the opportunity to peek behind the curtain of all the campaign management systems to see how smart or how dumb they are.
This complexity doesn't make it easy for advertisers to choose among the formulas, SEMs and technologies. It is interesting to read the latest Jupiter SEM Constellation Report (if you subscribe to Jupiter Research or can look over the shoulder of someone who does), just as interesting are the companies in the 2004 report that either chose not to participate in the 2006 report or perhaps were not selected by Jupiter to participate (Jupiter won’t say).
Regardless of what technology you choose to use to run your campaign, make sure that the math makes sense to you, for you, for your business and within your industry sector. Make sure you or the team using the technology is smart and understands your objectives. A smart Search Professional with the most rudimentary technology can probably beat the best technology implemented by a moron  |
Thanks for contributing, Kevin; sorry for my tardy reply, but I was out on paternity leave (boy #3
). A few comments on your post (quoted above):
"Some folks like to sell portfolios or a portfolio-style campaign management technique as mystical or magical. It's math."
You're right in saying that many firms claim portfolio management capabilities; I tried to thoroughly cover the history of how that came to be in my long post on this thread. The point I make is that Efficient Frontier (where I work) popularized the portfolio concept through effective stewardship of over $250M in ppc spend, and since then the portfolio notion has gained much currency in SEM, including the illegitimate usage of firms who would rather not lose business for lack of what EF has led advertisers to expect.
I also make a point that you seem to miss, which is that there are no indications whatsoever that any firm other than EF is applying portfolio algorithms to keywords. As you say, it is math, but not just a simple form of math as you imply; de-emphasizing the importance of portfolio algorithms is what we always hear from SEM's whose math is just that... simple. The fact remains, though - unless you're using algorithms to model keyword sets collectively and not just if-this-then-that rules that look at keyword data individually, you cannot possibly be aware of all the yield scenarios for that keyword set, let alone consistently find and implement the optimal set of choices.
People
You like to emphasize the value of people over math. I share that sentiment but can't disagree with you more when it comes to the real SEM world we live in, where shortages of trained online marketing account managers are acute. If I have as many people as you do, they're as smart (or smarter) than yours are and they know as much or more about the inner workings of the SE's systems as yours, then what's going to make the difference? Algorithms and data cross-keyword portfolio modelling, if you ask me. Your references to 'AI' and 'formulas' are intriguing, but I think it's high time SEM firms provide more than just vague intimations of math that's more than a glorified Excel spreadsheet. We've done that, and others should too, if they can.
Jupiter
Thank goodness you provided that link to the Jupiter report, without which I'm certain hundreds of advertisers wouldn't have known how to choose an SEM
. Really, though, anyone who bases their opinions of SEM firms on what Jupiter says obviously hasn’t any familiarity with how the technology research business works. Since getting a strong rating from Jupiter – or any other firm – involves alternately paying for their research, spending valuable time educating them (when I say 'educate' I use it in the Stalinist gulag sense of the term) and/or sponsoring conferences, webinars and websites, it's no surprise that the firms that fare well are those who play the research firm 'game'. Fortunately, most advertisers think of SEM firms like hedge funds – and good hedge funds spend zero effort marketing themselves, and 100% of their efforts getting strong returns for their clients.
"A smart Search Professional with the most rudimentary technology can probably beat the best technology implemented by a moron."
That's catchy, but in the ruthlessly efficient market we live in, I
certainly don't see any one SEM firm or advertiser with a lock on talent. If I can get you to admit that there are thousands of people (many of whom frequent this forum) who know every bit as much as you and yours about ppc, tactics, search engines and marketing, then we get back to the thorny issue of math and what math works best in search.
On that topic (which is what started this whole thread in the first place), I humbly submit that true, honest-to-goodness portfolio-style optimization of keyword sets is not a vague, subjective concept as you would have readers believe, but rather a set of algorithms and modelling techniques that you either have... or you don't.