Ellerton Whitney – Search Marketing

Google wants people to get the results regarding what they are looking for. This is part of their business model, and they believe it is why people keep coming back to Google. Thus, Google basically disallows you from showing up if your results are not deemed relevant enough. Sometimes you can pay a lot of money to show up when you are borderline irrelevant, but if you are totally irrelevant, you basically cannot show up.

Here is an example. Imagine you are a marketing manager trying to sell real maple syrup via the internet. You can run on maple syrup words and you will show up in Google results. For whatever reason, your company thinks that people who are interested in beef jerky will be very likely to buy maple syrup. For this reason, you decide to try to purchase beef jerky words. However, it turns out the people searching for beef jerky words don’t want to click on your maple syrup ads. Your “click through rate” heavily determines your “quality score” (aka relevancy). Thus, if people aren’t clicking, your quality goes down, and if your quality is too bad for a given word you won’t be showing up.

Thus, with Google, a lot of the time you are forced to be “passive.” Generally you are going to be unable to create new demand through search results. Traditional marketers froth at the mouth when thinking of ways to think outside the box, or ways they can conquest their competitors. In many instances, this isn’t possible with Google Search. For my automotive campaigns, I am able to bid on numerous manufacturer names because I offer that product. However, if I am trying to schlep Car XYZ to people searching for Car ABC, it becomes infinitely more difficult and frequently cost prohibitive. If you really want to try to create demand via internet marketing, I suggest taking a look at Google’s Contextual Network.

I haven’t written anything about SEM in a long time. This is big, though. Real big. It’s Google Instant.

Google Suggest has been out for years now, Google Instant takes this concept to a whole new level. Recall that Google Suggest would auto-populate a Firefox toolbar or a Google homepage search box with popular queries. Now Google Suggest will auto populate the entire search results page based off of the query a searcher is beginning to type. The common example now being used is that the results for “Weather” will show up when you start to type the letter “W.” This occurs whenever someone searches from Google.com or from a Google results page – you do not have to be logged into a Google account, and this is the new default setting (though it can be turned off if someone so desires).

My big issue with this is that Google is supposed to be a search engine. Some might say that this is helping searchers to search smarter. It is true, many people have not picked up how to search smart. However, Google is going to start eliminating the profits of the advertisers that butter their bread. Don’t bite the hand that feeds you, the old adage goes. With Google Instant, Google is going to shepherd people into the short-tail queries that create more profit for them. This will drive up the price of those short-tail terms as more people will be searching on them, and a competition increase will obviously ensue.

Not only will short tail terms become more competitive, lining Google’s pockets, but this will also begin to eliminate the long tail of search. For savvy marketers, the long tail of search is a great place to get high value conversions at a lower cost. With the elimination of the long tail of search, Google Instant has the potential to greatly impact the ROI of Google advertisers.

Don’t be evil, indeed. Google, the number one Search Engine? The number one Search Herder, I say.

As most junior search marketers, knows Google’s Broad Match is quite liberal in matching your keywords to search queries. Indeed, this is the intent of the match type, but sometimes it gets pretty crazy. Recently I had a scenario come up where the search queries that were mapping to the keywords did not seem logical. It was something like this:

Search query: car dealer sale
Keyword 1: car sale
Keyword 2: auto sale

Both keyword 1 and keyword 2 were on broad match. Logically, you would think the search query should map to keyword 1 because it has more words in common (2 out of 3 instead of 1 out of 3). I had a chat some folks in the know, and we got into something called “match length” – basically, a scoring system that Google uses to choose which of your own keywords will “win” to enter the SERP auction. We also got into how Google looks at the semantics of the keywords, and how that plays into match length. With the example, one would logically think keyword 1 would get +100 points (hypothetical) for car, and +100 for sale. In turn, Keyword 2 would only get 100 points, because it has only one word which actually matches. Points are assigned for other characteristics as well (see bad MS Paint illustration below)

In actuality, keyword 2 got selected over keyword 1. This is because the unique keywords were viewed as the exact same things. This has to do with how Google’s system “re-writes” keywords. Google re-wrote the word “auto” to “car” because the intent is so similar (they call them “related words”). Google verifies these re-writes internally by looking at the CTRs of the re-writes, and if the CTR is bad, it eventually verifies a bad re-write. Other ways Google re-writes are spell correction, and so called “query broadening.” Thus, each of the keywords appeared as if they were the same. So keyword 2 was selected due to factors other than the words themselves (geo-targeting, bid, quality score). In this case, keyword 2 was chosen because it had a higher ad rank (and ad rank is your quality score times your bid).

Some other details of how the paid search algorithms work were divulged to me. Some of them were pretty “duh,” such as the fact that exact match will always trump phrase and broad match, and other fairly common-sense nuggets. Exact match will always, always win, regardless of ad rank, quality score, or bid. So, if you bid your exact keywords down to a nickel, beware! However, a keyword in campaign that is more geo-targeted than another will get more “points” when considering which of your keywords to choose. If someone is sitting in Memphis, TN, and they perform a search with a query that could potentially map to multiple keywords in your account (all other things being equal), the likelihood of the keyword in a campaign geo-targeted to Memphis is greater a keyword in a nationally targeted campaign. I assume all of this was put into layman’s terms so that a search marketer such as myself can understand the basics of the algorithm, but you get the gist of things.

Google makes it very difficult to have precise control with Broad match. The best way to mitigate for the “efficiencies” Google creates is with many ad groups and proper negative keywords to ensure cross -coverage. This can become cumbersome quickly, however, so pick and choose your battles wisely!

Unless you are one of the lucky few managing some sort of brand PPC campaign that has incredibly difficult goals such as low Cost Per Click and high Click Through Rate, chances are you actually have to get some bang for your buck with your search engine marketing efforts. By that, I mean you want conversions — whether they be leads, sales, or otherwise.

When managing a pay per click campaign, it might seem intuitive to eliminate or pause keywords that are not driving conversions. This is not always the best idea, however, depending on what search engine you are optimizing in. For example, if you have the keyword ‘shoes’ on Broad match in Google, and ‘blue shoes’ on Broad match, keywords related to ‘blue’ and ‘shoes’ will be mated to your keyword ‘blue shoes’ (in some way, given that Broad match is overly Broad). Let’s say it is incredibly difficult to find blue shoes on your e-commerce site. You probably shouldn’t have been bidding on blue shoes in the first place, but for the sake of argument, assume you were bidding on this keyword. Because blue shoes are so hard to find, your CPA will be extremely high on the keyword blue shoes, so you want to get rid of it (some might lower the bid by 25 percent, or half, but let’s assume we are trying to get rid of a keyword altogether). To “get rid” of the keyword, most people would either delete it, or pause it. In Google, this is not the always the greatest idea. Unless you are adding the keyword ‘blue’ as a negative keyword (and let’s assume you do not want to do that because you also sell a brand of shoes called ‘Blue Magic’), your queries that were formerly matched to ‘blue shoes’ on broad match will now be mapped to ‘shoes’ on broad match, driving up the cost per sale on the broad match keyword ‘shoes’.

So what is the solution? According to Google, the LAST thing to occur in a keyword auction is the bid portion. So before the bid, Google will choose which keyword/match type to display an ad for, regardless of bid. So, instead of deleting or pausing your ‘blue shoes’ broad match keyword, you might consider lowering the bid to a penny. Effectively, this is “getting rid” of the keyword, but preserves the mapping of the non-ROI-yielding search queries to that bad keyword. No one will be clicking on your ads because your ads will not show up due to insufficient bid (or they will show up off the first page, and will receive no clicks). This is the best of both World’s — you have eliminated people searching on queries that do not yield ROI from seeing your ads, and you do not have to add any negative keywords.

If you are dealing with exact (not phrase) match keywords, it may be more acceptable to delete or pause a keyword. However, keep in mind that if you pause the exact match of a keyword, it may be mapped to another broad or phrase match keyword, so you may want to negative exact match it.

One other thing to note — as any experienced paid search marketer should know, Yahoo! keyword matching is relatively unsophisticated. Yahoo! matching works nothing like Google matching, as Yahoo! matching strives to get you the lowest Cost Per Click for a particular keyword. For example, Yahoo! matching will mate up the search query ‘flat screen television’ to ‘buy television’ if the keyword ‘buy television’ is in an ad group with a higher quality score. These keywords hypotheticals, but I have seen similar scenarios occur in Yahoo! where search queries are not mapped to what would seemingly be the most logical keyword. This has been explained to me by Yahoo! as trying to get advertisers the lowest CPC possible. So for Yahoo, there is really no rhyme or reason for what you should do with a keyword you want to eliminate, because if you lower the bid, it is hard to know what Yahoo! will map queries to that used to go to that keyword. Due to this complication, I usually just pause crummy keywords in Yahoo! if they do not convert… after I have tried all the possible methods to get the keyword to convert at an acceptable rate of course. If anyone has a better method, I would love to hear it!

It is commonplace for search engine marketers to drag and drop their campaigns from Google to Yahoo and MSN (and sometimes Ask). This approach needs to be treated with a grain of salt, however. It is important not to write off the other search engines, particularly Yahoo. There is a lot of excellent volume to be captured, depending upon the vertical you are targeting.

Some campaigns perform horribly in Yahoo and MSN. Indeed, it is sometimes hard to get anything out of MSN due to the low search volume. It is also easy to become frustrated with Yahoo and MSN bulksheets, especially if you do not have a ton of experience working with them. I have worked alongside search engine marketing managers who have completely written off Yahoo, and sometimes understandably so.

Truthfully, I have not always been the largest proponent of Yahoo’s search technology. However, in many cases, it is possible to get a lot of quality business or leads out of Yahoo — again, if you are in the right vertical. You might even get more than you are getting in Google! There are always exceptions to the rule, but it is often times the case that the Yahoo and MSN searchers are less savvy than the Google searchers. They also tend to be older and less educated (again, not always). It is possible to make logical conclusions about what campaigns should perform well in Yahoo and MSN, and those that should not. Without giving away every client I ever worked on, suffice it to say, I would think a Rolex e-commerce site would do poorly in Yahoo, and lead generation for The University of Phoenix could do quite well.

There are always exceptions to the rule, but it is often times the case that the Yahoo and MSN searchers are less savvy than the Google searchers. They also tend to be older and less educated (again, not always). It is possible to make logical conclusions about what campaigns should perform well in Yahoo and MSN, and those that should not. Without giving away every client I ever worked on, suffice it to say, I would think a Rolex e-commerce site would do poorly in Yahoo, and lead generation for The University of Phoenix could do quite well.

I found it very interesting when I logged into each search engine (today is the Martin Luther King, Jr. Holiday). Google has a very slick looking image of Dr. King on a handsome background. Yahoo has something that looks like Monopoly Money. Well this is not direct evidence of the different audiences (and the brains behind each operation), it works out to being a funny comparison of the two search engines.

In conclusion, if you are running paid search and you have the budget for all of the engines, make sure to give Yahoo and MSN a whirl. They aren’t always going to work well for what works in Google, and sometimes they won’t work well at all. But it is worth dealing with those clunky bulksheets and the Panama UI to find out, because there is still a lot of quality traffic to be captured.

Best Practices are Best Practices for a reason. One of the issues with Search Engine Marketing today, however, is that people are catching on to the best practices. When it comes to Pay Per Click, you only have so many characters to work with. There are still ways to break through the clutter, however, even when everybody’s Pay Per Click ads are stuffed with keywords, benefit driven, and use keyword insertion.
1. Take advantage of the 40 character ad titles in Yahoo.

No advertisers do this. Despite the difficulty of using the Panama interface and the cumbersome nature of the Yahoo Bulksheets, Yahoo can still yield excellent results — many times you can achieve CPL/CPA’s lower than with Google, and at a substantial volume. Most advertisers just dump their Google exports into Yahoo. But look at this screenshot — Position 1’s title really stands out from the crowd, and I am sure it has a much higher click through rate for it!

2. Take advantage of new Google features

I might be shooting myself in the foot by putting these out there, but they aren’t a secret — people just aren’t using them (yet). One unique new feature is to display the address of your business in your ad (if you appear in Google’s top 3 positions). This is only good for small business or something with an actual location, however. For e-commerce, take advantage of Google Checkout icons, which boost click-through-rates as well. Always keep your eyes out for Google’s latest and greatest options in AdWords.

3. Use Unique Characters

When writing ad copies, use characters that stand out from the crowd, especially with Branded Campaigns. Parentheses. {Official Site}. Or use a reserved trademark symbol, or a copyright symbol.

4. Employ a display URL strategy

Many SEM campaigns aren’t properly using display URLs. Associate your display URLs with the keywords in your ad group to boost your click through rates.
5. Don’t forget the best practices!

Finally, don’t abandon the best practices! I have come across those that want to be out of the box for the sake of being out of the box. If you do this, you will most likely fail. Best practices work. You can do all of the aforementioned, but usually those crazy out-of-the-box ideas from the HIPPO aren’t going to work. But look for ways to properly differentiate your ads. Tell a story. Make it simple. Do something different, but keep to the best practices as well.

Updated 9/9/2010, Originally Posted 12/21/2008

The Google Content Network has greatly evolved. I retract previous statements that the Google Content Network sucks, because it really doesn’t anymore. Yes, there are still some Made for AdSense farms out there, but some of them get good traffic through SEO and produce valuable leads or conversions. I still maintain that for lead generation Content Network leads may be of a lower quality than Search leads, but the Content Network is a far cry from where it was a few years ago.

The key to optimizing the Content Network today is treating websites like keywords, and also targeting your content to relevant keywords on these sites. Google’s placement reports will tell you where your content network placements are showing up. If you target the entire network, this is a good way to get a baseline of where you can show up on the Content Network with the keywords you have populated your Content Campaigns with. As your Placement Reports fill up with placements, add all of the websites as placements, and bid them up or down as necessary based off of ROI (or eliminate some altogether).

The Content Network might not work for everyone, and certain verticals are better than others to be sure. The targeting is still not a perfect science, but it’s much better than it used to be and is far more manageable than it used to be with the introduction of keyword-targeted placements that can be bidded individually. I used to be a naysayer, but if you are overlooking the Content Network, it is time to re-evaluate.