Antiquated Internet Creates Geographic Problems
Twinkling banners, frame-based web pages, and cutting-edge ISDN innovations; the 90’s were the Wild West of the Internet. By the late 90’s, fiber was being laid in major cities at a rapid pace and the 1 MHz processor barrier was finally broken. Underpinning all of this was the one threat that could undermine it all, the Y2K bug.
History tells us that the Y2K bug was a tad overblown, and the undoing of the fiber revolution was the DotCom Bubble Burst. Companies like Google have purchased a great deal of those old fiber cables and begun finishing the job that was started at the time.
But the sad fact remains, the U.S. Internet infrastructure hasn’t advanced much since then. And it is still a challenge all companies are dealing with.
A Little History
At one point, the AOL was the dominant ISP of the United States, and during that time, all of their servers were hosted in Virginia. This really didn’t matter for the longest time, as services such as Yahoo! were nationwide to begin with. You bought keywords, but you couldn’t choose a geography.
The Trouble Deepens
When Google decided the future was geographical targeting, they had to find a solution to the problem. Since geographical targeting for most advertising platforms is based on the user’s IP address, this meant that a huge portion of the US Internet traffic appeared to be in Virginia. This required a lot of work by advertisers to work around, and it meant a lot of advertising dollars would stay offline.
The advertisers had two strategies. In the first one, they target the entire US. They could then choose every possible geographical term possible and hope there wasn’t too much overlap with cities like Dallas, Texas and Dallas, Indiana. Just take a look at the 10 Most Common City Names in the US. It was easy to suddenly see tons of traffic from irrelevant users.
The second strategy was selecting more generic terms while targeting only the desired geography. That worked okay, but it omitted a significant portion of the user base.
Needless to say, both strategies were employed by many advertisers.
Local & Personalized Intent
Google realized that in order to coax more dollars online, they were going to have to get smarter. The smarter method required adding sophistication to their SERPS at two levels. Google needed to understand local intent at on a per-query basis, and Google needed to know their customers.
It is important to understand how users search geographically. Many users trust that Google will simply provide the most relevant results. They may search “plumbers” to find results for a local plumbing service. But many users want to provide more assurance that the results they are shown will be appropriate. These users may be more likely to add the city, city and state, or simply the zip code to their search.
A quick search for “plumbing” can provide a myriad of possible results, from DIY videos and articles, to services, to a possible listing on the etymology of the word ‘plumbing’. If a user is looking for a plumber, the services listings are the only relevant results. Many advertisers may not want to be shown on such a broad query. Overall, this is hardly an ideal query or keyword for any party.
A quick search for “plumbing services” suddenly provides a lot better results for users looking for plumbing services. But users that have chosen to geo-concatenate their query may have an another way of tipping off their intent. Try searching “[your city] plumbers”, and the results are also very similar to the “plumbing services” results. Taking this a bit further, adding the state abbreviation is guaranteed to have map listing results appear, as in “your city, state] plumbers”.
Just imagine the reasons for such a variety. I live across the metroplex from where I work. I speak in other cities. Isn’t it possible that I’m searching for a business that is across the metroplex from where I’m at? Isn’t it also possible that I’m searching for a business where I am traveling to or where I have been recently?
The Crazy Chaos of Geo-Signals
This can get crazier when you know how many users are knowingly or unknowingly bouncing out via a proxy. Military units overseas could be routed via a regional Internet proxy and back to the US.
The geography signals are not quite as simple as that when you factor in GPS and wireless providers for mobile devices, which in-turn may bounce through via a user’s Google profile to each of their devices.
Furthermore, every browser has a digital fingerprint which is difficult to cover. Then again, how many users use the same computer. Imagine the empty-nesters that have their college age or older kids come home for the holidays and login to their own profiles.
How does Google handle it all?
Well, Google decided to continue watching the user’s geography, or at least the geography of the user’s ISP. But they would also ‘guess’ the user’s intent based on their search history and activity. So, a user in Las Vegas could reasonably receive ads for a Miami plumber if they planning an East Coast vacation in the Florida Keys recently.
Just imagine the implications for the inexperience advertiser. They have to trust that Google is determining the intent and geography signals correctly. Needless to say, this can create a lot of cost overruns. The less relevant results may also be creating problems with the Quality Score.
While this may seem like a money-grab by Google, modern AdWords geography targeting is an enormous improvement over the old targeting of simply using the IP’s. The real strategies to deal with very local businesses involves advanced methods of the two old school geographical targeting methods I described earlier. I’ll describe those methods in a future blog article.
Now that you understand the way Google targets better, go make some money!