In one of my earlier postings attempting to detail my read on how specifically Google so thoroughly has beaten Yahoo! in the business of search, I touched on the power of Google’s broad match technology. Today we look again on this topic.

First a little background. “Broad match” is one of the three primary “match types” (the other two are exact match and phrase match) that are used to create Google advertising campaigns. Broad match is very effective in allowing basic keywords to be matched to a wide variety of user search queries, giving advertisers easy access to reach a bigger audience. Sounds great, right?

Well, historically it has been pretty darn effective, especially for less sophisticated advertisers that do not have the ability to generate super-comprehensive keyword lists for their campaigns. Originally, broad match worked by helping to match a keyword to related keywords, fixing misspellings, pluralization and other helpful devices. So your keyword “los angeles lawyer” would be triggered by user queries such as “best los angeles lawyer”, “los angeles lawyers”, “lawyers in los angeles”, and other closely related keywords. Last year, Google starting expanding the scope of broad match, using it’s massive knowledge of how users search the web. Continuing the previous example, this meant Google may link you to synonyms such as “attorney” or abbreviations such as “LA”. This is intriguing because a search for “LA attorney” would trigger your keyword “los angeles lawyer” even though at this point there were no overlapping words from the user query to your search campaign. However, this doesn’t seem too bad because “LA attorney” and “los angeles lawyer” seem like fairly synonymous intents.

In the last several months, however, broad match has spun out of control. Google claims that “the algorithm is constantly learning” and that no evil is intended, but we are starting to see very, very scary leaps that broad match is making and are moving quickly to adapt to this new beast. For example, we have seen the keyword “buy rugs” come up under the search “reno carpets”, “purses” pull up “luggage” and “los angeles lawyer” come up under “california law”. This, quite simply, is wrong. A rug is not the same as a carpet, purses are not luggage, and Los Angeles and California are related but quite clearly distinctly unique keywords.

Our wonderful Google rep echoed the danger of broad match, indicating that using broad match is something that needs to be done with caution: “If you sell coca-cola, you may be matched to diet coke, and maybe as well orange juice because all three are liquids.” I have universally positive things to say about Google as a search engine, advertising platform and company. However, for a company that pledges to “do no evil”, this technology is toeing the line of unacceptability. If I am a new advertiser and sell coca-cola, and Google shows my ads to people searching for orange juice, these are wasted dollars, and frankly, I am going to make a stink because this is not what I signed up for! What makes it all the more confusing is that if I look in my AdWords account, it will tell me that I received the click for my keyword “coca-cola” even though the click may have been from the user searching for orange juice. Unless I have the knowledge to add negative keywords or evaluate the new Search Query Report or analyze my web site logs, I will have no idea that Google is mis-spending my money.

What makes this especially scary to me is that broad-match is not some special type of keyword that advertisers can opt into in order to get more exposure. It is the default keyword type, so unless you specifically employ other keyword match-types, broad match is what you get.

So what does this all mean?

• Google makes more money. I am not suggesting that expanding the power of broad match to this scope was something done purely to make more money, but regardless of intent, the bottom line is that more ads that are matched to more search queries makes Google more money.
• Advertisers that are unsophisticated or inattentive are either going to spend more money than they could be for the same results, or are going to perform so poorly they look for professional campaign management or quit outright.
• Advertisers that are sophisticated are going to have to adapt and find ways to build campaigns that is comprehensive enough so that they can use the other match-types and don’t have to “take the bad with the good” of broad match. At Wpromote this is a major part of the technology we are building out to combat this; finding ways to build back in the traffic lost if you stop using Google’s broad-match technology (which as a rule of thumb seems to provide about 50% of a campaign’s traffic. If you have a campaign running in exact match and broad match, and pause the broad match, your traffic drops approximately in half).
• Google’s users suffer. This is a key point for Google to keep in mind. The underlying claim across all of Google’s initiatives is that they are geared towards enhancing the user experience. Showing overly-broad-matched ads to users is absolutely a negative for the user experience. If I am searching for coca-cola and you show me orange juice, I begin to lose faith that Google can give me high quality results. If the argument of hurting advertisers isn’t enough to motivate Google to change, then certainly the argument that as a byproduct of this aggressive broad-match you are degrading the user experience. And if users keep searching, and finding less and less targeted results in those “little blue boxes” (the sponsored link ads that generate Google well over $10 billion a year), they are going to stop clicking on them, and if enough people start doing this, users will find other places to find the information that they seek.

Save this Post!

Hide Sites

Last week, we stumbled upon what appears to be a quietly rolled out development in the Google Adwords ranking algorithm.

If you are one of the thousands of Adwords advertisers that regularly do Google searches to confirm your ad’s positioning on key search terms, you may notice some very strange things going on, as our most involved clients and acute account executives found in the last week.

Several days ago, I was checking our ad positioning on a small handful of top spending keywords for our internal marketing efforts, such as “online advertising”, “search engine advertising” and “ppc management”. Expecting to see us in the top 1-3 results in the banner as usual, I was shocked to find us in considerably lower positioning, 7-8 in most cases. Quickly logging into our accounts, I saw that our average position that day was what I would expect, around 1-2 in all of the tested keywords. Why would I be seeing such dramatically different results?

While I was waiting to hear back from one of our terrific Google Agency Team reps, I tried a few things, including clearing my cookies. I tried the searches again, and we are back at the top where we should have been.

What does this mean? Basically, if you repeatedly do the same searches on Google — within the same cookie session or while logged onto a Google account — without clicking on the links coming up top, Google assumes that you aren’t getting search results in the ads that interest you, and show you different ordered results. Essentially, this means that you are getting results that are not an accurate reflection of the normal ad-rank algorithm, which takes into account your bid, your click-thru rate and a number of other, less weighted factors.

To the normal Google user, this will go entirely unnoticed. For PPC advertisers and their obsessive managers (I for one catch myself doing test searches in Google without really thinking about it, completely in auto-pilot sometimes) this means that you may find yourself confused about as positioning that is different from what the rest of the world sees.

What to do? You can clear your cookies for one, that seems to remedy the situation. If you do not want to clear your cookies (which results in websites logging you at and a myriad of other annoyances) you can use an alternate browser from your preferred program for test searching, and regularly clear your cookies on that one. Or you can use the Google-recommended route, which is to perform all test searchesusing Google’s Ad Preview Tool. The benefit of using this tool is that by appending the URL with the appropriate codes (as explained here), you can see how users in any specific zip code, region, country or other geo-targeting option will see results. This is excellent for evaluating other markets and doing test searches for campaigns targeted to areas outside your own.

In my several conversations with Google regarding this issue, there has been significant confusion and frustration among advertisers, but time will tell whether there is any change in the way this new development operates.

UPDATE: I am curious as others test this to find out if others see what we seem to be noticing. Specifically, the URL that we are “looking for” in these test searches seems to be the one that is coming up lower in the searches. In other words, we dot not seem to be seeing a general jumbling of all the first-page ad results. Is Google tracking other activity, linking URL’s in Google Adwords accounts or looking at other data? At this point we need more data and it is too early to tell. If you have noticed anything like this, let us know; we are eager to learn more about this subtle, intriguing Google update.

Save this Post!

Hide Sites

First, an apology. After an exciting start, I fell completely off of the blogging wagon, caught up with a major office move and a subsequent hiring frenzy along with a myriad of other tasks. Great explanation, but poor excuse.

This entry marks my official return and pledge to stay up to date. I have a massive list of topics and news to cover, and look forward to the coming days, weeks and months.

Yours Truly,

Mike

Save this Post!

Hide Sites