Sample Page Zero Advisor Newsletter

VOL. 5 NO. 4 | NOVEMBER 2007
Biz Models That Are Getting Google Slapped

By Andrew Goodman

Download This Sample Issue [PDF]

Welcome to another fresh update from the world of paid search.

I just got in from a great trip to Stockholm (the SMX conference) -- had a terrific time consorting with the Scandinavians. One thing that I found really nice was a visit from Gillian Muessig of SEOmoz, coming over from the “organic side of the wall”. She sat in on a session on Paid Search Fundamentals (as moderator and main presenter, I was given 25 minutes, and gulp, went on for 33 minutes, which meant I couldn’t say anything as Mel Carson of Microsoft and Stefan Waldeck of Yahoo blew past their allotted times), and asked a number of questions to liven things up in the Q&A. After all these years, at the conferences it still seems like we need to redouble our efforts to “break down those walls” between the organic and paid tracks, lest we get stuck in our own narrow siloes.

One of the hot topics around the dinner table (other than how many pairs of shoes Tor Crockatt owns, that is) was the release of Microsoft’s new analytics platform, named Gatineau. This is an exciting development, because it gives Google a run for its money and offers the prospect of deeper demographic data layered over web analytics. Plus, it reminds me of the place I used to ski (Gatineau hills) as a boy growing up – in fact, that’s why it was named Gatineau, because it was hatched at an Ottawa-based research centre or something like that. Dixon Jones declared it to be a “stupid name,” but then cheekily noted that the first two letters are “GA,” so there is some competitive logic there. (Don’t they own GoToast now?) According to Mel, Microsoft “encourages” people to install two or more analytics services rather than trying to convince them to lose their existing solution in favor of Microsoft’s. So there you have it.

OK, enough insider banter and obscure geek jokes.

Before we get to the main course, just a note that here at Page Zero, we are accepting one more early preview client for our Landing Page Juice multivariate landing page testing service before it rolls out at the full retail price. (The preview price is about 50% off the full price.) Time is tight as our schedule fills up, so if you’d like to initiate a conversion improvement project before year end, book now. Please contact Cory Kleinschmidt at cory [at] page-zero.com and indicate: 1-3 key pages on your site you’d like to improve. Criteria: you must have some decent click volume (any source is fine, paid, unpaid, direct referral, etc.) or the test will take forever. Our methodology involves an overarching set of theoretical considerations, as well as specific creative inputs (design, layout, copy), and of course a methodological approach to proper testing so that the experiment is done right.

If you’d rather schedule something for the new year, that can make more sense because conversion rates are less volatile post-holiday. Our price for 2008 won’t be rock-bottom, but still good value.

Contents:

0. Conversion Testing Is Hotter Than Ever (Book Now)

1. The Types of Biz Models That Are Getting Google Slapped, (or Downright Boxed About the Ears)

2. Mastering Panama: It’s Time to Get Yahoo Rocking

3. Google Trying To Behaviorally Target Ad Serving – Ouch, Say Us

4. PPC Assurance – Get a Handle on Invalid Clicks and Scary Patterns

5. How’s Your Year-Over-Year?

6. Seasonal Strategy: Google Discloses Conversion Rates, CPA Trends in Holiday Period

Sign Up: Be Notified When You Can Subscribe

1. Stuff the AdBot Usually Hates, from a Quality Standpoint

Recently, I wrote an article on the latest, latest in Google’s ongoing quest to tweak its modern “quality score” algorithm – the one that decides where your ad ranks on the page… or whether it’s displayed at all.

http://searchengineland.com/070925-140955.php

More recently I followed up with Nick Fox, Director of Product Management in the ads division at Google, to gain some added insights.

Nick wanted to clarify that Google isn’t specifically targeting business models outlined in their FAQ page; rather, that list is just a helpful overview of some of the types of sites that in Google’s experience to this point, “are more likely to get into trouble areas.” So publishing the list was an attempt at “additional transparency” on Google’s part. To be clear, then, there has been no marked change in policies to affect, say, “get rich quick” sites. It’s just a guideline to explain that these kinds of sites more often than not run into quality score problems.

I’m aware that quality score rules are shaped by human input, including input from users. In the case of users, I asked how they are conveying preferences to Google. Fox cited:

• Writing into Google (using the feedback form re: dissatisfaction with a page of search results)
• How they behave [an example would be leaving the site immediately and returning to Google]
• Email trickling into various Google people
• Phoning Google!

OK.

So in a more structured sense, Google also runs “human evaluation programs” on an ongoing basis.

I wanted to drill down on what Google meant by “travel aggregators” and why they run into trouble frequently according to the FAQ. I got some interesting feedback from Fox on this one. Basically, he is suggesting that they are often “not really travel aggregators” and are fronts for some pesky type of affiliate or arbitrage activity. In other words, Google will go after the “sketchy ones.”

Reiterating the point that the bulk of the evaluation is done formulaically in automated fashion, he did allude to the human element in Google’s evaluations. This appears to work in two ways.

First, if someone feels that they have been unfairly targeted, or if Google staff notice that a site seems to be unfairly caught in the quality score web, they can “go through it and take corrective action,” in essence manually resetting the quality score on the account or on some keywords.

In the opposite case, editors can also take direct negative action to manually worsen quality score if they discover that quality scores assigned are better than they should have been.

Without going into further extensive analysis, I hope this clarifies the situation somewhat. Many of the quality score problems specifically related to website or landing page quality are in fact associated with business models that Google has explicitly attempted to go after, because users don’t like these things. In other cases, there may be user experience issues on sites that make them somewhat questionable, in a kind of “yellow flag” zone. In these cases some combination of human and automated means are used to set poor quality scores, making it much more expensive for advertisers to appear on their chosen keywords. In rare cases, it may be possible to get some clarification from Google and in essence “appeal” the situation. Google appears to be still grappling with the pros and cons of telling you specifically what is wrong. Whoever the “bad guys” are, Google doesn’t want them knowing exactly what they’re looking for, especially when it comes to the workings of the
automated scoring algorithm. Let’s hope that’s not you.

Remember, the other kind of quality score is not site related – it relates to CTR’s, account history, other relevancy factors, and bids. That’s the core ranking formula that affects everyone in the auction, whether they are “bad” or “good.”

^ Back to Contents ^

2. Info-Torial: Is Your Yahoo Campaign a Shipwreck?

While in Stockholm, I spent a few hours with Chris Sherman walking about the city, including a stop at the great shipwreck museum (probably not the best description, but a display of a huge warship that sank in the 17th century, salvaged and restored to perfection after its discovery in 1959).

Now, if your Yahoo campaign looks like a shipwreck… hmmm… nah, too salesy… :)

But still, there is a Swedish tie-in to this story. You may be aware that Yahoo recently decided to roll out its new quality scoring (Google-like) formula here in Sweden, about a week ago. “Full Panama” is not available around the world, but only in a few selected countries. The lucky Swedes will be able to see what we Canadians saw as Panama rolled out shortly after it did in the U.S. – the better ads bubbling up to the top as the smartest advertisers connect with consumers and get rewarded with higher ad positions at lower cost.

The opportunity is clear even if you’re in the U.S. The majority of advertisers are taking poor advantage of the new features in Panama, and as a result, are leaving money on the table.

Mona Elesseily’s very timely Mastering Panama: A Report on Yahoo Search Marketing’s New Platform is available to Page Zero Advisor subscribers at a special discount price, 20% off the cover price. Even if you get just two new ideas from this book, it’ll pay for itself in a few days! Grab it now at the Page Zero store – it’s a print-on-demand hard cover that will look attractive next to your coffee-stained copy of Winning Results with Google AdWords ;) .

Buy now: http://www.cafepress.com/pagezero.157497830
(discount price of 20% off the $89 cover price)

3. Google’s Controversial Ad Serving Continuing to Give Headaches

Try as we might to be masters of our own domains, control over AdWords campaigns sometimes seems as elusive as ever.

You may recall that it has become dicey to rely too heavily on broad matching, because of the unpredictable way Google sometimes serves up ads using “expanded broad matching.” The antidote to this, typically, is to employ negative matches (if you’re energetic), or to be a little more clever in your bid strategy and matching options strategy.

There’s another development afoot: Google is experimenting on so-called “behaviorally-driven” ad serving where they may, in some cases, show an ad based on a *past* search, next to search results for the *current* search. Let’s say someone types in “ford bronco” and a little while later types “shoes”. Your ad for “Ford Bronco brake pads” could conceivably come up next to some query down the line after that initial Ford related search.

The specifics give me a headache, I admit. Although there are suggested workarounds in this great article by Mike Churchill, it probably shouldn’t be incumbent on you to be trying to disable Google matching features through obscure unpublished methods. Until Google comes clean and decides whether they are or are not going to document their “behaviorally-driven experiments,” (along with a proper “opt out” capability), the only thing to do is to sit tight, and again, focus on better bidding strategy, and caution with matching options.

Bottom-line-wise, Google felt this experiment was material enough to disclose it in their recent quarterly earnings call. It might have been nice, though, if they’d spoken with more advertisers about it. We were the guinea pigs.

To me, it’s a transparency question. Do all or most advertisers have an equal understanding of how the auction works? If we come to the table with certain assumptions, is it fair of Google to game and tweak those assumptions to the benefit of Google’s profitability, without fully disclosing the changes being made to the workings of the auction? If your ad wasn’t “good enough” to show on Page 1 on the first ford bronco related query, then shouldn’t that be your cue to bid differently or to expand your account? Google’s trying to “help you” guess at who is searching for you, but you can be sure they won’t be helping you to do that with your lower-bid keywords. There’s a difference between helping little old ladies across the street, and installing a catapult that throws all passersby across the street so we can get there “better”. And then charges you on your card for the service.

Hey Google, we’ll walk.

Hats off to Mike Churchill for a thorough look into this issue.

4. Pay Only for Real Clicks!

If you’ve been a subscriber for awhile, you probably know that I don’t often endorse third-party products.

You may also know that I’ve been skeptical at times of the claims of “click fraud auditing software” vendors.

Of the available aids I've tried, I think that PPC Assurance, by Enquisite, is the most elegant and accurate solution. Best of all, you can now check it out totally risk-free with a free trial offer.

Here's where to go for that:

http://ppcassurance.enquisite.com/couponcode.html

Then, on signup, use coupon code PPC-GOOD-1109

(If you don't use the code, you won't get the free trial. Richard has promised me he'll go in and extend the trial offer for Page Zero Advisor subscribers only... the general public are only getting a very short trial or discount, not a real free trial. If you don't want to continue with the service, all you need to do is cancel by the deadline, and you pay nothing.)

I was involved in some beta tests of the product. What I saw emerge was a simple system that shows graphically which clicks you need to be alarmed about. Richard Zwicky of Enquisite takes a bit of a different view of so-called click fraud than many vendors. He and his development team have divided up the world into red, yellow, green, and gray. Red clicks are clicks that fall outside the terms of service you agreed to when you paid for clicks, and that you got charged for. Yellow clicks were outside the parameters you chose, but the search engine didn’t charge you for them. Green (the vast majority) were clicks that fell inside the terms of service. Gray are unknown clicks – associated with the difficulty of tracking those users (perhaps they have javascript disabled, or there are problems with your website).

Some other click fraud vendors are calling half or more of those yellow and gray clicks “fraud,” which grossly inflates the true extent of the problem.

A couple of things emerged in testing PPC Assurance.

First, the product does a great job of showing you the sheer volume of weird clicks, and details about them. These would include a lot of bad clicks from countries like China. But these aren’t all click fraud, because Google or Yahoo didn’t charge you for all of them. It’s interesting to understand what’s going on just by drilling down in the reports.

Second, we saw very different patterns emerging in “amateurish” accounts as opposed to highly optimized accounts. Most professional campaign managers (like my friend Scott here at Page Zero) do a meticulous job of testing, building out the tail, and bidding to ROI criteria. The very act of properly organizing and refining a campaign over time (Scott’s been working on this test case for over two years) makes almost all of the click fraud disappear… like magic. (As you may know, I implied as much in my very early versions of the Google AdWords Handbook and said so again in the Winning Results book. It’s good to see some graphic proof of this assertion, confirmed by the developer of third party click auditing software.) While this is heartening, there are still a few blips of red to look at.

Richard Zwicky informs me that some other accounts he’s seen – less professionally managed ones – have a lot of red in them. Whether you choose to mend your account management ways or go directly to Google to request a refund, PPC Assurance arms you with unbiased data to help guide your strategy going forward – with no chip on its shoulder.

Go ahead and give it a spin – risk-free, and free of charge - for a precious few days. :)

Here's the full info on how to get the FREE trial (the offer that will allow you to cancel anytime before the extended deadline Richard has promised us).

Start your signup at:

http://ppcassurance.enquisite.com/couponcode.html

On the payment information page (specifically the credit card entry screen) enter the coupon code "PPC-GOOD-1109". The purchase process will only accept that code or an empty coupon box.

^ Back to Contents ^

5. How’s Your Year-Over-Year?

Apples to apples aren’t always pleasant comparisons in paid search, because costs have risen so much in the past couple of years. It’s no mistake that Google has earned record revenue quarter after quarter. They’re monetizing the dickens out of that search engine of theirs, and bargains are getting harder and harder to find.

But don’t let yourself fall into the temptation of panicking over month-to-month changes in performance over two and three month periods. While it’s never pleasant to see things drop off, you might find it more instructive to do a little archival research (“Raiders of the Lost ROI”) and try to dig into what were the sources of your success, and your shortcomings, 12 and 24 months ago. I don’t know of any business that doesn’t look closely at the year to year comparisons, yet somehow in paid search things move so fast that people don’t want to turn the page back.

You might just run across some interesting discoveries. On one campaign I discovered several things by looking at its “last year” numbers. The one that stood out was that a large number of promising ad groups had dropped in position. While the client (and later, us) was content to bid to ROI criteria, the increased competition in that space meant that this tactic led to volume drops on the best words… from positions of 2 and 3, to 5, 6, and 7. This client didn’t have a particularly restrictive CPA ceiling, so how ridiculous was it to let a bunch of core keyword groups get knocked down to positions 6 and 7! Leads immediately jumped, albeit with some increase in CPA, when we bit the bullet and went back at positions 2-3.

A number of other interesting discoveries were made by looking at old data. The fact of an October “lull” being some kind of crisis, in light of that, was pure superstition. That lull was just minor seasonality, a victim of too much continuity and incrementalism in strategy. A deeper assessment was needed.

^ Back to Contents ^

6. You Have Been Warned: Don’t Screw Up This Year’s Seasonality

Google has a Chief Economist. This shouldn’t surprise anyone, given that their annual output exceeds that of many small nations.

On the Inside AdWords blog, we’re treated to a chart showing the aggregate conversion rates, CPC’s, and CPA’s as the holiday season swings into gear. Google can provide this because a goodly percentage of AdWords advertisers have conversion tracking turned on. It’s a little like Google Trends in Google Labs in that the charts don’t give you absolute figures, only a relative sense of changes over time. Still, it’s very useful information.

What you do with the information is of course up to you. But it’s pretty clear that advertisers are making the following two mistakes:

(1) When retail conversion rates begin to spike up, giving everybody pretty much their lowest CPA’s (cost per acquisition, a generic team being used to cover cost per order or any other significant conversion event typically measured by advertisers) of the year, advertisers are too slow to increase their budgets and bids, so they miss out on customers at this key time. They eventually react and climb onboard.

(2) Then, they’re tardy in ramping their bids and budgets down as sales really dry up starting around the 24th. So a nice chunk of the profitability just flies out the window because somebody forgot to slash bids for the dead period following Christmas. Yikes. Making two mistakes this big could cut into your profit picture for the entire year. Pay close attention to conversion rates starting in mid-November.

Thanks for reading! I'll see some of you at SES Chicago next month! And Mona will catch you at Pubcon in Las Vegas.

Cheers,

Andrew Goodman
www.page-zero.com

"Most of the people who try to sell you search engine optimization are just ripping you off. Inside Andrew Goodman's special reports, though, I've found real, common-sense approaches that ought to work for just about anyone. Teach a man to fish..."

Seth Godin
Author, Meatball Sundae